It is important, at the threshold, to be clear regarding the species of "economic analysis" to be used. The method employed in the [*1699] following pages compares alternative legal rules on the basis of their capacity to promote "economic efficiency," which is defined as "that allocation of resources which could not be improved in the sense that a further change would not so improve the condition of those who gained by it that they could compensate those who lost from it and still be better off than before." n194 More specifically, the method takes as given the existing distribution of income and wealth in American society n195 and the monetary values n196 placed by each member of the society on different combinations of goods, services, and states of affairs. It then asks what legal rule governing unauthorized uses of copyrighted materials would yield the combination of production and dissemination of works of the intellect that is most efficient in the sense just described.

The ensuing discussion of the analytical paths a lawmaker might take in answering that question is divided into five sections. The first sets forth the economic rationale for the copyright system as a whole and shows the relationship between that rationale and the fair use doctrine. The second sketches a hypothetical fair use problem, simplifies it with a number of assumptions, and then outlines a method that a well-informed and underworked judge might use to identify its [*1700] most efficient solution. The third pauses to consider the uses and lessons of the proposed method. The fourth discusses how the method might be refined to make it applicable to complex cases. The fifth distills from that more sophisticated approach a set of guidelines federal courts might practically employ.
 
 

A. Determining Optimal Levels of Copyright Protection

From an economist's standpoint, the trouble with works of the intellect is that they are "public goods." n197 Unlike most goods and services, they can be used and enjoyed by unlimited numbers of persons without being "used up." It is thus difficult to deny access to such works to persons who have not paid for the right to enjoy them. n198 These conditions create a risk that inventions and works of art that would be worth more to consumers than the costs of creating them will not be created because the monetary incentives for doing so are inadequate. Laws forbidding members of the public from copying or making other use of intellectual products without the permission of their creators are designed in part to eliminate this source of economic inefficiency. By granting inventors and artists a type of property right in their products, the law induces creative persons to develop and exercise their talents and thereby avoids the underproduction of useful ideas and original forms of expression. n199

Unfortunately, this solution may foster economic inefficiency of a different sort. Granting an artist or inventor a property right in his creation may make him a monopolist, giving rise to familiar economic distortions. To the extent that consumers regard other intellectual products as only imperfect substitutes for a particular copyrighted or patented work, the holder of the copyright or patent will confront a downward sloping demand curve for the right of access to his work. Under such conditions, if he wishes to maximize his profits, he will continue granting access to his work only up to the point where the marginal revenue he reaps from affording access to an additional [*1701] consumer equals the marginal cost n200 -- while at the same time charging a price substantially higher than his marginal cost. n201

Adoption of the foregoing strategy by a copyright or patent holder will have two economic consequences. First, he will reap a monopoly profit; in other words, money that would have remained in the pockets [*1702] of consumers, had the work been priced at the level at which the marginal cost of producing it equalled the demand for it, will now go into the pocket of the copyright or patent holder. n202 Second, consumers who value the work at more than its marginal cost but less than its monopoly price will not buy it. n203 The former effect is usually thought to have no predictable impact on allocative efficiency. n204 The latter, however, results in a "deadweight loss," measured by the total of the consumer surplus that would have been reaped by the excluded consumers and the producer surplus that would have been reaped by the copyright owner had he sold the work to them. n205

The degrees of market power enjoyed by different copyright holders n206 -- and thus the severity of the dangers just recounted -- vary [*1703] considerably. At one extreme are copyrighted works that consumers consider irreplaceable. n207 At the opposite extreme are works for which (in the eyes of consumers) there are readily available, nearly perfect substitutes. n208 Most types of copyrighted material lie somewhere between these poles. n209

In summary, to avoid underproduction of original works, it is necessary to empower the creators of such works to charge fees for the privilege of using them, but granting the creators that right causes monopoly losses, which vary between types of copyrighted works. The task of a lawmaker who wishes to maximize efficiency, therefore, is to determine, with respect to each type of intellectual product, the combination of entitlements that would result in economic gains that exceed by the maximum amount the attendant efficiency losses. n210 Roughly speaking, the "gains" associated with a given combination of rights are the value to consumers of those intellectual products that would not have been generated were creators not accorded those rights. More precisely, they equal the present value of what consumers would be willing to pay in the future for the works whose creation is induced by the rights, minus the present value of the costs that [*1704] would be incurred in producing them. The associated "losses" are the deadweight costs discussed above. n211

A lawmaker setting up a copyright regime "from scratch" could pick from a large array of possible entitlements when considering which combination would work best in the sense just described. n212 The range of options available to a judge, when exercising his authority under section 107 of the statute, would seem to be much narrower. Other provisions of the Copyright Act seem to settle most of the relevant issues, thus consigning the judge to the job of making minor adjustments to the overall system of entitlements. n213 A review of the language and legislative history of section 107, however, suggests that the judge's sphere of action is not so limited. The mandate of the provision is expansive: the "factors" courts are required to consider in assessing fair use claims are deliberately vague; n214 the list of factors is incomplete; n215 and the judiciary is encouraged to modify the doctrine in response to changing conditions and technology. n216 Moreover, the potential reach of the doctrine is broad: a judicial determination that a given use of copyrighted material is "fair" relieves the defendant of all liability, "notwithstanding" the statutory provisions vesting in a copyright owner certain exclusive rights. n217 Thus, if the package of entitlements created by the provisions of the Copyright Act other than section 107 enabled the creators of a type of intellectual work to collect monopoly profits from consumers in greater than optimal amounts, a judge could use the fair use doctrine to chip away at that package until it approximated the most efficient combination.

In short, at least with regard to works of some sorts, economic analysis of the fair use doctrine is nearly coterminus with economic [*1705] analysis of the copyright system as a whole. n218 We now turn to how that analysis should be conducted.
 
 

B. Fair Use as Efficient Use: The Basic Approach

Two months ago, Plaintiff, a famous mystery writer, published a hardcover edition of a volume of detective stories, More Adventures of Mr. Holmes. The current retail price of her book is $20. Defendant, a college professor, recently distributed photocopies of one of the stories in More Adventures to the twenty students in his seminar, "The Art of the Short Story," charging the students only the cost of the copying. Hoping to deter other teachers from engaging in similar conduct, Plaintiff has brought an infringement action against Defendant, seeking compensatory or statutory damages. n219 Defendant argues that his activity constituted a "fair use." The judge n220 wishes to decide the case in a way that promotes economic efficiency. What should he do?

The complexity of the problem necessitates approaching it in stages. This section begins the job by adding a large number of simplifying assumptions to the hypothetical question and showing how the judge might proceed in such an imaginary world. The section after next introduces a series of refinements to that basic method in the hope of developing a set of tools a real judge might use on a real case. n221

[*1706] The judge's first task is to expand his frame of reference. For reasons that will become evident later, n222 it will not suffice for the judge to concern himself solely with Defendant's conduct. To decide the case, he must identify and compare (on efficiency grounds) all of the uses of copyrighted detective stories n223 that might be considered fair or unfair. In other words, he must determine the universe of activities vis-a-vis detective stories that would violate the Copyright Act if not excused by section 107.

Assume that the judge concludes that there exist only five such uses of copyrighted detective stories. Conveniently, More Adventures is about to become the target of an example of each one of those activities, Plaintiff has learned of these threats and has sought to enjoin all of them, and the judge himself will hear all five suits. The activities in addition to Defendant's copying are:

(i) Pirate Publishing Company is about to release a nearly identical hardcover edition of More Adventures, which will retail (at least when it first appears) for $15.

(ii) Bestseller Book Club will soon make available to its members a paperback edition of More Adventures at a price of $4.

(iii) Barker Brothers, the dominant manufacturer of board games in the United States, has developed a game called "Detection," the variations of which correspond closely to the plots of the stories in Plaintiff's book. n224 The game will soon be available in toy and hobby stores for $10.

(iv) The New York Times plans soon to publish a review of More Adventures written by Calvin Critic. Critic's principal arguments are that Plaintiff's style is more turgid than in her previous sets of stories, but that the plots of the new stories are more clever. To substantiate these points, Critic quotes several passages from More Adventures and from Plaintiff's earlier works.

The judge's job, then, is to decide which (if any) of the activities should be deemed fair and which (if any) should not. Assume, finally, that the conditions for taking on such a project are ideal: the judge [*1707] has no other cases on his docket; he has unlimited research capabilities; and he knows that his decision will not be reversed and will be accepted in all other jurisdictions. How should he proceed?

His first move should be to rank the five uses of Plaintiff's stories in order of the relative benefits and costs of legitimating them. To do that, he must determine for each putatively infringing activity what hereinafter will be called its "incentive/loss ratio." The numerator of this fraction is the net reward a detective writer of Plaintiff's stature n225 could reap if the activity in question were forbidden and the writer could therefore demand a fee from any person who wished to engage in the activity. x The denominator is the efficiency loss that would result from empowering the writer to set the terms on which others could engage in the activity. n227 The purpose of the ratio is to provide the judge with a preliminary indication of the net economic benefits of according authors the entitlement in question.

Set forth below are speculative accounts of how determination of the incentive/loss ratio might proceed in each of Plaintiff's five suits.

(i) The judge's analysis of the case of Pirate Publishing Company reveals the following. Many readers of detective stories regard Plaintiff as one of the best contemporary writers in the field and would much rather read her stories than stories by other authors. Plaintiff, because of her prominence, was able to negotiate a favorable contract with her publisher, pursuant to which the publisher agreed to abide by Plaintiff's instructions with regard to the manufacture and marketing of the book and to retain a modest portion of the purchase price of each book. x Plaintiff's estimate of the demand for More Adventures was accurate, and the retail price of $20 was shrewdly designed to provide her with the maximum monopoly profit possible [*1708] on the hardcover edition n229 of the book. Plaintiff has enjoyed those monopoly profits for the past two months and would like to continue to do so. Pirate selected a price of $15 for its competing edition in hopes of garnering some of Plaintiff's monopoly profits. That price, however, is highly unstable. If Pirate's conduct were held to be noninfringing, Plaintiff would lower the price of her book to match Pirate's price. Pirate would, in turn, retaliate with an even lower price. n230 Continued competition would quickly drive the retail price down to the cost of manufacturing and marketing a copy. n231

On the basis of those findings and projections, the judge could estimate with comparative ease the incentive/loss ratio associated with the use of detective stories exemplified by Pirate's behavior. If Pirate's activity were held to be fair, price competition of the sort just described would soon eliminate both Plaintiff's monopoly profits and the concomitant deadweight losses. In the simple case represented graphically in the margin, n232 [*1709] the ratio of the former to the latter would be 2 to I. n233

(ii) The judge turns next to the behavior of Bestseller Book Club and discovers the following. The reason that, as yet, Plaintiff has released only a hardcover edition of More Adventures is that she is engaged in price discrimination. n234 Potential purchasers of her book vary in their eagerness to read it; in general, the more eager a purchaser, the more he is willing to pay for a copy. Knowing this, Plaintiff plans to delay release of a paperback edition (which will retail for $4) for another six months, thereby forcing avid readers who do not mind soft covers to purchase a hardcover edition, on which Plaintiff's profit margin is high. n235 By offering its members a cheap paperback edition, Bestseller would force Plaintiff to release her own paperback edition earlier than she had planned, thereby depriving her of the profits from the sale of the hardcover edition.

Based on these findings, the judge concludes that, for two reasons, prohibition of Bestseller's activity would result in a substantially higher incentive/loss ratio than would prohibition of Pirate's conduct. First, price discrimination enables a monopolist to increase his profits without materially increasing the deadweight losses associated with his profitmaking. n236 Consequently, the ratio between the reward to [*1710] Plaintiff and the concomitant monopoly losses caused by protecting Plaintiff's price discrimination scheme would substantially exceed the 2:1 ratio associated with the sort of "ordinary" monopoly pricing described above. Second, even though the competitive price for a paperback edition would be less than $4, Plaintiff probably will not engage in price competition with Bestseller but will allow Bestseller to reap higher-than-competitive profits from its members, enabling Plaintiff to retain higher-than-competitive profits from nonmembers. n237 As compared to a situation in which the only paperback edition on the market were Plaintiff's (selling for $4), this equilibrium would result in a diminution of Plaintiff's profits with only a slight reduction in monopoly losses. The judge concludes, therefore, that the incentive/loss ratio associated with Bestseller's activity is very high.

(iii) The judge's study of the case of Barker Brothers reveals the following. Barker's size enables it to produce and sell board games much more cheaply than its competitors and, therefore, to engage in monopoly pricing of "Detection." n238 The manufacture and sale of "Detection" is unlikely to have any effect on either the sale of More Adventures or Plaintiff's reputation. Accordingly, if Barker were forbidden to make and sell the game without Plaintiff's permission, the two parties would negotiate an agreement, under which Barker would pay Plaintiff a royalty for each game it sold. n239 From Barker's standpoint, that royalty would increase the cost of each game. Barker [*1711] would respond by increasing the price at which it sells the game. n240 The results of this series of actions and reactions would be: Barker's monopoly profits from the sale of the game would decline; n241 deadweight [*1712] losses associated with game sales would increase because fewer potential consumers of the game would be able to buy it; n242 and Plaintiff would keep a portion of the monopoly profits from the sale of the game. n243 The ratio the judge must ascertain is that between Plaintiff's new-found income and the increase in deadweight losses. n244 On the assumptions embodied in Figure 3 in the margin, n245 the ratio would be 1.76, somewhat less than the ratio associated with Pirate's behavior.

(iv) Next on the judge's agenda is the behavior of Calvin Critic. The judge's investigation reveals that Calvin's comments are accurate (in the sense that most readers of More Adventures would agree with Calvin's judgments after reading the book) and that the net effect of publication of the review would be to reduce slightly the demand for the book and consequently Plaintiff's income. The efficiency losses associated with preventing that diminution in demand would be very large, however, because forbidding publication of Critic's comments would prevent many consumers from making informed decisions to purchase or not to purchase the book and thereby prevent them from maximizing their welfare. n246 Consequently, the incentive/loss ratio associated with prohibiting Critic's activity is very low. n247

(v) The judge turns finally to the behavior of Defendant. He discovers that the practice (in which Defendant and many similarly situated professors are now engaged) of making copies of individual copyrighted stories for distribution to their students affects Plaintiff's [*1713] revenues in three distinct ways. First, some students who otherwise would have bought copies of More Adventures find their desire or curiosity satisfied by reading a photocopy of a single story and decide not to purchase the book, thereby depriving Plaintiff of income she might otherwise have earned. Second, the interest of some students who would have been unaware of the existence or quality of Plaintiff's work is piqued by reading a single story, and they purchase copies of More Adventures or of Plaintiff's other books, thereby increasing Plaintiff's income. Third, the prices Plaintiff can exact from publishers producing anthologies of short stories for use in courses of the sort taught by Defendant are less than they would be if professors had more incentive to assign such anthologies.

Keeping these impacts in mind, the judge next tries to determine what would happen if he were to declare Defendant's conduct unlawful. Research persuades him that professors in Defendant's position would respond in three ways. n248 Some would abandon their current syllabi and would instead assign (and require their students to purchase) anthologies containing one of Plaintiff's tales. The result, the judge determines, would be to enhance Plaintiff's revenues modestly. n249 Adoption of this approach would have two economic costs, one of which would be substantial. First, by increasing the total cost of the courses in question, it would deter a few students from enrolling. x Second, and much more important, by preventing teachers from using the combination of readings that best matches their conceptions of the appropriate content of their courses, this reaction would impair the quality of the education provided to the students not so deterred, thereby reducing their welfare without enhancing the welfare of anyone else. In sum, the incentive/loss ratio associated with this response would be fairly low.

Another group of professors would respond by abandoning their current syllabi and assigning anthologies that did not include one by [*1714] Plaintiff. This strategy would result in economic losses of the same sort and magnitude as those just discussed. n251 Because it would not increase Plaintiff's royalties, the impact of this course of action on Plaintiff's revenues would depend on the relative magnitude of the substitution and exposure effects. The judge ascertains that the benefit to Plaintiff caused by the elimination of the substitution effect and the injury to Plaintiff caused by the elimination of the exposure effect would approximately offset each other, and therefore the ratio associated with this response would be extremely low.

The third and last group of professors would not alter their syllabi but instead would request their college libraries to purchase and place "on reserve" a few copies of More Adventures and would require students to read the assigned story in the library. The consequent increase in Plaintiff's royalties (from the sale of copies of her book to the libraries) would enhance her net income significantly. n252 The increased inconvenience to some students x and the reduced quality of the education afforded the remainder n254 would result in some economic costs, but they would not be as substantial as those associated with the first two responses. The judge therefore concludes that the incentive/loss ratio associated with this response would be relatively high.

To assign an aggregate ratio to Defendant's conduct, the judge would have to multiply each of the incentive/loss ratios derived in the preceding three paragraphs by the percentage of professors x who would adopt the course of action in question, and then add the products. Following this procedure, he concludes that the composite ratio is larger than that associated with Critic's conduct but smaller than that associated with Barker's.

At long last, the judge is in a position to rank the five uses. The activity with the highest incentive/loss ratio is the publication of a paperback edition contemplated by Bestseller. Next comes Pirate's competing hardback edition, then Barker's game, then Defendant's copying, and finally Critic's review.

[*1715] Now what? The judge's ultimate objective is to select the combination of entitlements that will maximize net efficiency. n256 By adding up the denominators of the fractions he has just derived, the judge can determine the efficiency losses associated with any given set of entitlements. To ascertain the efficiency gains, however, the judge must perform one last set of calculations. Remember that the point of allowing Plaintiff to demand from each consumer of his book more than the marginal cost of producing an additional copy is to induce both Plaintiff and other actual and potential writers of detective stories to write more and better stories in the future. The source of the efficiency gains referred to above is the increased consumer satisfaction that results when detective-story readers have access to a better menu of stories. The judge thus must estimate the present value of the change in detective-story writing that would be caused by each increase in the aggregate reward offered to successful writers like Plaintiff's. n257 To make that estimate, the judge must take into account a host of variables: the strength of motives other than thirst for financial remuneration that impel people to enter and ply the trade of detective-story writing, n258 the other occupations available to detective-story writers and the incomes associated with them, the ability of an incipient writer accurately to predict the likelihood that he would succeed in the business, the risk aversion of the typical writer, and the discount rate.

Our judge diligently makes the necessary computations and determines the efficiency gains associated with each level of total income [*1716] that Plaintiff could be permitted to reap from the sale and usage of her book. n259 His findings are plotted on the following graph. n260

[*1717] To ascertain the most efficient interpretation of the fair use doctrine in this context, the judge need only identify the point in the series of putatively infringing uses where the difference between aggregate efficiency gains and aggregate efficiency losses is greatest. On the facts as hypothesized in Figure 4 above, that spot is the activity of Barker Brothers. By holding Barker's behavior and all uses to its left (Bestseller's and Pirate's) unfair and all uses to its right (Defendant's and Critic's) fair, the judge will, to the limit of his ability, maximize allocative efficiency. n261

C. Taking Stock

Is all this really necessary? Must a judge, in order to evaluate the fairness of a defendant's activity, assess and rank all other conceivable uses of the plaintiff's work? In two extreme situations, the answer is no. First, with regard to some sorts of copyrighted works, it may be that to forbid even a single use would result in efficiency losses that would exceed gains. Note that, in the hypothetical case, Plaintiff was able, without any legal protection, to reap two months of monopoly profits from the sale of her hardback edition. That income might be sufficient to induce many prospective writers to try their hands at detective stories. Even the substantial increase in Plaintiff's income caused by preventing Bestseller from releasing a paperback edition [*1718] might not provide enough additional incentive to offset the efficiency losses such a prohibition would produce. n262 With regard to copyrighted works of other sorts, it may be that to permit even a single putatively infringing use would incur economic costs that exceeded gains. For example, it might be that detective story writers are extraordinarily sensitive to fluctuations in their anticipated incomes, and that, as a result, the gain to consumers of forbidding book reviews altogether would exceed the efficiency losses that such a ban would cause. n263 In situations of these two types, it clearly is unnecessary to rank uses. Every use of a copyrighted work of the first sort should be ruled fair; every use of a copyrighted work of the second sort should be ruled unfair.

Except in these special contexts, however, maximizing efficiency through interpretation of the fair use doctrine unfortunately requires the kind of comprehensive, comparative analysis sketched in Section B, for two reasons. First, the relationship between the aggregate rewards available to artists of a given kind and their production of works valued by consumers is almost certainly not linear; therefore, the efficiency gains associated with forbidding a given activity depend on what other uses are permitted and proscribed. Second, if the incentive/loss ratio of any activity held to be fair exceeds the ratio of any activity held to be unfair, the optimal solution has been missed.

If such a comparative analysis must be employed in most cases, is not economic analysis in this doctrinal context hopelessly impracticable? The discussion of the highly stylized case presented in section B was complex enough. If we removed the simplifying assumptions, limited the judge's investigatory power, and burdened him with other cases, it would be ludicrous, surely, to ask him to undertake an inquiry like the one outlined above. Perhaps. It is hard to imagine a judge making even rough guesses at some of the figures critical to the calculus. Especially implausible are the notions that he would be able, or willing, n264 to ascertain the universe of ways in which copyrights [*1719] in the type of work before him are susceptible of being infringed and that he could estimate the effects of different levels of remuneration on the future production of works of that sort.

Even so, the analysis may have considerable value. The assumptions used in section B were not wildly unrealistic. Some of the conclusions reached by the hypothetical judge may survive transition to the real world. Moreover, a simplified version of the procedure might enable a court at least to increase allocative efficiency, if not to maximize it. Before we can derive those lessons, however, we must take into account some important considerations thus far excluded from the discussion. n265

D. Complications

1. Categories of Works and Uses. -- In the hypothetical case, the judge was concerned only with the production of "detective stories" and knew that there existed precisely five distinct ways of infringing the copyright in such stories. A real case would not be so nicely [*1720] packaged. Instead, the judge would be obliged to classify (i) the plaintiff's work and (ii) the various arguably illegal uses of it.

The first of these tasks would be harder than one might expect. From an economic standpoint, it is important that authors be able to predict with some confidence what uses of their works will be considered fair and what uses will be considered unfair; otherwise, the expected income from creating those works will be uncertain, and they will not be optimally induced to develop and exercise their talents. n266 Differentiating types of copyrighted works for the purpose of applying the fair use doctrine thus creates some inefficiency, because each division decreases artists' ability to predict how their creations will be classified and, therefore, what they will be worth. On the other hand, some differentiation seems essential. Copyright law currently covers a wide array of original works, ranging from books to films to computer software programs to advertisements. Both the responsiveness of creators of those various sorts of works to different levels and kinds of incentives and the economic impacts of different infringing uses of such works vary dramatically. Consequently, to lump all intellectual products together when attempting an analysis of the sort sketched in section B would be senseless. In short, categorizing copyrighted works for the purpose of determining what uses are fair would promote allocative efficiency, but creating too many subdivisions would reduce efficiency.

How should a judge determine how many lines to draw and where to draw them? Assume, for example, that he must decide a case involving a parody of a Broadway musical. Should the rule he prescribes cover all plays, all musicals, all Broadway musicals, or some other category?

To make an efficient choice, the judge should estimate three figures. First, he should determine how much of an efficiency gain, caused by differentiating the monetary incentives for creating different types of intellectual products, would result from moving from each level of specificity to the next. For any of a variety of reasons, the optimal combination of entitlements for musicals may be different from the optimal combination for other sorts of plays. n267 Lumping them together would therefore result in a less-than-ideal system of incentives for playwrights of all sorts. The judge should ascertain how much less than ideal. Second, the judge should determine how [*1721] much of an efficiency loss, caused by reducing playwrights' ability to predict their incomes, would result from each subdivision. Suppose, for example, that differentiating musicals from other plays resulted in sharply different combinations of entitlements and income levels for writers of the two sorts of works, n268 but left a substantial number of potential playwrights uncertain as to whether the plays they contemplated writing would ultimately be classified as musicals or nonmusicals. The effect would be to sacrifice much of the efficiency gain described above by reducing the willingness of risk-averse playwrights to write works whose status was unclear. Third and finally, the judge should calculate how each additional differentiation would affect the number of cases in which the parties disputed the classification of the copyrighted work in question, and what the litigation costs generated by those controversies would be.

If he were able to get this far, the judge's task would be easy. He would subdivide a category of copyrighted works only if the first figure were greater than the sum of the second and third.

To determine how finely to slice the set of infringing uses, the judge would have to undertake a similar analysis. Assume that, in the case just described, the judge settles on "musicals" as the appropriate category and then conducts an analysis of the sort sketched in section B to determine whether parody is a fair use. He concludes that parodies fall to the right of the optimal point and, therefore, should be considered fair. It then occurs to him to differentiate parodies that use just enough material from a musical to "conjure up" the original in the minds of the audience from parodies that use more material than necessary to achieve that effect. n269 He modifies his analysis to incorporate this distinction and concludes that the optimal point lies between the two types of parody. Should he distinguish between them in deciding this case and in announcing a rule to govern future cases?

The answer depends on the relationship between three figures: (i) the efficiency gains associated with improving the incentive system by distinguishing among parodies; (ii) the efficiency losses caused by deterring risk-averse parodists from making parodies the statuts of which would be uncertain; and (iii) the costs of litigating cases in which the [*1722] parties disputed whether more material was used than necessary. If (i) exceeded (ii) + (iii), he should make the distinction; if not, not.

2. Stare Decisis. -- To this point, we have been talking as if the judge, in deciding a fair use case, were writing on a blank slate. Given the current state of the doctrine, it may make sense to approach the job of reconstruction in just that spirit. However, new decisions would soon begin to fill in the slate. From an economic standpoint, how should the accumulation of decisions affect the treatment of new cases?

Assume that, after a certain date, each judge confronted with a fair use defense conscientiously undertakes an analysis of the sort described in section B and writes an opinion indicating which uses of the type of copyrighted material before him should be considered fair and which uses should be considered unfair. n270 So, in the case of Plaintiff v. Defendant, Judge #1 rules that issuing unauthorized paperback or hardcover editions of sets of detective stories and selling games based upon such stories are all unfair, whereas copying detective stories for educational purposes and quoting portions of them in critical reviews are fair. A year later, a case involving educational copying of detective stories comes before Judge #2. Judge #2 conducts another evaluation of the problem, relying in part on empirical data concerning the production of detective stories collected during the preceding year, and concludes that the optimal point lies between educational copying and criticism. Should he repudiate Judge #1's ruling?

If Judge #2 were certain of his findings, could costlessly inform all prospective detective-story writers of the new rule, and could persuade them that the rule would never change again, the answer would surely be yes. None of these assumptions, however, is likely to be true. The judge therefore should compare the present value of the efficiency gains that he anticipates would result from altering the incentive system (multiplied by the probability that he is right) with the costs of spreading the word to affected parties plus the efficiency losses associated with undermining the confidence of artists and incipient artists of all sorts that the legal system now in force will still be [*1723] in force when they are in a position to make some money from their creations. n271 Under these conditions, the new rule would have to be substantially better than the old to justify the change. n272

3. Property Rules or Liability Rules? n273 -- We have assumed, to this point, that a judge called upon to assess a particular use of copyrighted material has only two choices: he can declare it unfair, thereby forcing the user to bargain with the copyright owner for permission to engage in it; or he can declare it fair, thereby empowering the user to engage in it without compensating the owner. In principle, there exists an intermediate option: the judge can declare the use fair if and only if the user pays the copyright owner a fee set by the judge.

From an economic standpoint, a judge's ability to employ a liability rule of the sort just described would have two advantages. First, in some cases it would enable him, by setting the price of engaging in a particular use at a level below the monopoly price that would be charged by the copyright owner, to increase dramatically the ratio of reward to monopoly loss associated with that entitlement. n274 Such a maneuver would reduce the income the copyright [*1724] owner reaped from the entitlement, n275 so the judge would need to grant the owner more entitlements than would be required if he employed property rules exclusively. The net result of such a general reform of the doctrine would be an enhancement of economic efficiency.

Second, in situations in which transaction costs would prevent a user from purchasing the right to engage in an activity if it were held unfair, the availability of a liability rule might increase a judge's choices. Reconsider, for example, the case of copying detective stories for educational purposes considered in section B. Our hypothetical judge concluded that, if teachers were not permitted to photocopy stories, they would either assign their students anthologies or place books "on reserve," and that the economic costs generated by those [*1725] responses would be substantial, rendering the incentive/loss ratio associated with forbidding the photocopying low. If the judge had the option of declaring such copying fair on the condition that the copyright owner receive from each student a royalty of, say, $.50 per story, the situation could change considerably. Most teachers might respond by assigning the same materials in the same form they had assigned before the fee was imposed. Although the result would be to increase somewhat the price of the course materials, thereby deterring a few students from enrolling, n276 the major source of economic loss -- the diminution of the value of short-story courses to the students who do enroll -- would be eliminated. The ratio of reward n277 to efficiency loss n278 associated with such a regime might well be quite high (especially if the size of the fee were selected shrewdly) -- perhaps high enough to alter the sequence of entitlements. n279

The foregoing analysis suggests that it is sensible for Congress, when modulating the entitlements of copyright owners in particular fields, to employ compulsory licensing systems, which permit specified uses of copyrighted works on the condition the users pay fees set by an administrative tribunal. In a few contexts, Congress has already [*1726] availed itself of this option. n280 It ought to do so more often. n281

The circumstances in which a judge, construing the fair use doctrine, can and should employ such liability rules, however, are likely to be more limited. Although a court's remedial authority under the Copyright Act probably encompasses the power to forbid a particular use until the user pays a royalty, n282 that authority almost certainly does not include the power to order the creation (and funding) of an administrative agency capable of monitoring a compulsory licensing system. The court itself, therefore, would be obliged to determine an appropriate schedule of fees and to alter that schedule periodically in response to changing economic conditions. Leaving aside the question whether a court would be willing to undertake a project of that order, in many contexts the associated litigation costs would exceed the potential economic gains. n283 Finally, the court's ruling would only [*1727] bind the parties to the suit in which the order was announced. Similarly situated parties would have an incentive to try to secure more favorable arrangements from other courts. Such duplicative litigation would compound the costs of the remedy. n284

In short, the efficient response to a suit implicating the fair use doctrine occasionally may be to employ a liability rule rather than the usual property rule. Sony may have been such a case. n285 However, the majority of cases probably will not lend themselves to such a solution.

4. Private Licensing Arrangements. -- The question whether, if a use were held unfair, the copyright owner and user might negotiate a licensing agreement arose in two phases of the "basic approach" presented in section B -- the discussion of the deal Plaintiff and Barker Brothers might arrange n286 and the discussion of the possibility of an efficiency-enhancing contract between Plaintiff and Defendant. n287 This subsection refines those analyses.

(a) Sophisticated Deals. -- Barker's dominance in the board-game market puts it in a position to pay Plaintiff a substantial royalty for the use of More Adventures. Section B hypothesized that, if Barker's unauthorized use of the book were held to be unfair, Plaintiff and Barker would negotiate a contract that would require Barker to pay Plaintiff a flat fee for each game manufactured and sold. n288 Such a contract is not, however, the only possible outcome of their negotiation. Indeed, if Plaintiff and Barker are sophisticated bargainers (or if they are advised by sophisticated lawyersn289 ), they are unlikely to reach such an agreement. Instead, they will devise some way of keeping the retail price of the game at $10 n290 and dividing the monopoly profits generated by that pricing strategy. n291

[*1728] A contract of the sort just described would fare very differently in the fair use calculus than would the arrangement discussed in section B. If the parties agreed not to alter the retail price of "Detection," a ruling that Barker's activity is unfair would not increase the efficiency losses already caused by its monopoly pricing scheme. The denominator of the fraction associated with Barker's use would therefore be only the cost of negotiating the licensing agreement -- a relatively small figure. The size of the numerator would be difficult to predict, n292 but most likely it would be much larger than the denominator. The incentive/loss ratio, consequently, would be very large -- probably large enough to require that Barker's activity be shifted to the left in the sequence of entitlements. n293

In sum, when assessing the economic impact of declaring unfair a derivative use of copyrighted material by a party in a position to engage in monopoly pricing of the derivative product, it is critical to predict accurately the content of the licensing agreement that would be precipitated by such a declaration. Generally speaking, the more sophisticated the parties, the more likely they are to make a deal that benefits the copyright owner without materially harming the public, and therefore the more willing a court should be to classify the use as unfair.

(b) Blanket Licenses. -- In situations in which many artists produce copyrighted works of a particular kind and many users are willing to pay for access to those works, the transaction costs that impede the negotiation of licensing agreements between individual artists and users might be overcome through the establishment of a blanket licensing organization. Such an organization might arise without governmental intervention in at least two ways. First, the artists might form an association to which they cede their rights to license their works. The association would then negotiate agreements with individual users and distribute the revenues to its members. Second, the users might form an association, which would negotiate agreements with individual artists. n294

The advantage of such organizations is that, by enabling artists to augment their revenues without incurring the efficiency costs caused [*1729] either by banning the putatively infringing activity altogether n295 or by administering a judicial compulsory licensing system, n296 they would dramatically improve the incentive/loss ratio associated with the activity at issue. n297 So, in the case of Plaintiff v. Defendant, discussed above, the judge, before locating Defendant's conduct in the sequence of arguably illegal activities, should assess the likelihood that holding educational copying of detective stories unfair would prompt authors or teachers to form such an organization, n298 and should modify the ratio associated with Defendant's conduct accordingly.

5. Synergy and Progress. -- The analysis thus far has implicitly treated creativity as an individualistic phenomenon. A good detective story, it has been assumed, is created out of whole cloth by a gifted individual writer. Thus, the way to improve the menu of detective stories available to consumers is simply to increase, through adjustments in the incentive system, the number of gifted writers who decide to exercise their talents rather than become lawyers. Creativity, however, is often -- perhaps always -- a more social phenomenon than this scenario suggests. Every writer, composer, and filmmaker draws on the work of his predecessors when creating something new, n299 and most are stimulated by the ongoing work of their contemporaries.

Sensitivity to this characteristic of artistic creation should prompt a judge to watch for two possibilities when estimating the efficiency gains and losses associated with a particular combination of entitlements. First, [*1730] when an art form is in its infancy, adjusting the compensation system so as to increase the number of artists engaged in developing it might substantially increase the quality and variety of products available to consumers in the near future because each participant in the venture would build upon and stimulate the work of all the other participants to an unusual degree. n300 As opportunities for synergy of this sort diminish, courts should reduce the copyright protection for such works by expanding the set of uses of them deemed fair. n301

Second, a judge should hesitate before declaring transformative or productive uses of copyrighted material unfair. In many contexts, an artist's ability to engage in creative work depends on his freedom to make use of other copyrighted works in ways that might be considered infringements. A parodist, to be successful, must mimic an original work at least enough to make the object of his ridicule recognizable to his audience. The designer of a computer operating system can generate a better system more rapidly if he is permitted to incorporate portions of other copyrighted systems. n302 Proscribing unauthorized activities of these kinds risks diminishing the stock of intellectual creations available for popular consumption. n303

[*1731] This insight can be made more concrete by returning once again to the case of Barker's use of Plaintiff's book. n304 To simplify the discussion of that case in section B it was assumed that neither the number nor the quality of board games based upon copyrighted stories would be impaired by a rule requiring developers of such games to purchase the right to use the stories. n305 That assumption may be unrealistic. The ability to invent interesting board games is a rare talent, and the willingness of persons possessing that talent to develop and exercise it is likely to be affected by their expectations of monetary rewards. Thus, if Barker were required to pay Plaintiff a fee and consequently to forgo a portion of the monopoly profits it otherwise would make on the sale of "Detection," the compensation it paid creators of such games would probably decline, and fewer or inferior games would be produced. The result would be an efficiency loss analogous to the efficiency loss caused by undercompensating detective-story writers like Plaintiff. That figure would have to be added to the denominator of the ratio associated with Barker's unauthorized use of Plaintiff's book, thereby reducing the size of the fraction and perhaps altering the ranking of the five activities enough to compel redetermination of the optimal package of entitlements. n306

The general point is that, when calculating the incentive/loss ratio associated with a creative use of copyrighted material, a judge must be sure to include in the denominator the present value of the consumer surplus that would be sacrificed if persons who engaged in the use were obliged to compensate the copyright owner and, as a result, produced fewer or poorer derivative intellectual products. n307

[*1732] 6. Frustrating Popular Expectations. -- The large majority of fair use cases involve issues about which most Americans are either ignorant or indifferent. n308 In unusual circumstances, however, there may exist something approaching a popular consensus regarding the fairness of a particular activity. The best example is the Sony case. By the time the Supreme Court rendered its decision, much of the population seems to have concluded that it was perfectly appropriate behavior to record copyrighted programs broadcast over the public airwaves. n309

Judges who desire to maximize allocative efficiency should be watchful for such situations and, when they occur, should be loath to issue decisions inconsistent with the popular view. The reason is that otherwise they risk undermining people's confidence that they can predict the content of the laws that affect their lives. Most Americans regularly make major decisions (for example, to lease an apartment, to buy or sell a car), the rationality of which depends heavily on the content of legal rules, with only the vaguest idea what those rules are. We act this way partly because we believe that, by and large, the law is just, and that, by consulting our sense of justice, we can guess with reasonable accuracy what the rules are and will be. Each time we are informed of a significant legal rule inconsistent with our sense of justice, that confidence is shaken. A person's loss of confidence of this sort will prompt him to respond in one of two ways, each of which will entail economic costs. He will either spend time he would not otherwise have spent learning what the relevant legal rules are before commencing a project, n310 or he will decide not to undertake the project at all. n311 The more widespread the upset expectation, the more common these responses will be and the larger the associated economic losses will be. To be sure, even a notorious violation of a popular assumption regarding the content of a legal rule may impair only marginally the willingness of Americans to engage in economic and social activity without investigating their legal rights, but the enormous aggregate economic losses that would result from [*1733] even a modest reduction of that willingness should be a cause for concern.

7. Shaping Preferences. -- In the discussion to this point, popular tastes for different sorts of intellectual products have been taken as given. More specifically, it has been assumed that a consumer's desires for books, movies, software programs, etc., depend neither upon his access to such goods nor upon behavior in which he or other members of the society might engage. n312 However, as "choice theorists" recently have begun to point out, n313 those assumptions are unrealistic; our tastes are affected in countless ways by our actions. n314 Our desires for some activities are intensified by engaging in them -- either because they are physically or psychologically addictive, like using heroin or jogging, n315 or simply because they provide pleasures that increase with familiarity, like playing chess. n316 Desires to own certain objects or enjoy certain rights are enhanced rather than diminished by indulgence; I am likely for a variety of reasons n317 to place a higher value on a bottle of wine or on the right to an odor-free environment if I am already entitled to it than if I am offered the chance to purchase [*1734] it. n318 We also frequently immerse ourselves in communities -- such as the Armed Forces, law school, and country clubs -- that we know will alter our personalities and tastes.

As the last example suggests, not only do our actions shape our preferences, but we often choose between alternative courses of action in order to shape our preferences. n319 In circumstances in which we worry about our capacities to control the evolution of our tastes individually, we frequently use systems of collective control -- most notably the legal system -- to manage our preferences. n320

In debates over the extent to which government should support the arts, the malleability of popular tastes has long been recognized. Much of the continuing controversy in that area concerns whether government officials may legitimately influence people's desires and, if so, how they should exercise that power. n321 Such questions have not hitherto been raised in arguments concerning the appropriate content of the fair use doctrine, but brief reflection makes plain both their relevance and the challenge they present to an economic analysis of the field.

Consider the following case. After conscientiously examining the various ways in which the copyright in a play n322 might be infringed, a judge concludes that the incentive/loss ratio associated with classifying a parody of a play as infringement would be slightly lower than the ratio associated with classifying a board game based upon the [*1735] play as infringement and that the optimal point lies between those two activities. He is therefore inclined to rule that making a parody is a fair use whereas making a board game is not. It then occurs to him that, if other courts follow his lead, his ruling may affect not just the availability and price of parodies and board games, but also popular tastes for them. Further empirical investigation convinces him that, if he reverses the order of the two entitlements and declares parody unfair and derivative board games fair, people will soon begin to lose interest in parody for several reasons. Few playwrights will grant parodists permission (even for a fee) to make light of their plays. n323 As a result, the supply of parodies will drop substantially, the price n324 of parodies will rise, and fewer people will see them. Parody is a subtle art form, appreciation of which depends in part on frequent exposure to it. Depleting the stock of parodies will therefore diminish popular desire for and enjoyment of parody. In addition, a psychological phenomenon that has come to be known as the "sour grapes effect" n325 will prompt parody aficionados to persuade themselves that they did not really want to see the parodies that they are unable to view. n326 Finally, that the law frowns on parody will diminish popular taste for it. n327 The diminution in demand that results from this combination of responses will, in turn, sharply reduce the monopoly losses associated with declaring parody unfair, n328 and will therefore increase the incentive/loss ratio associated with parody, n329 [*1736] making it higher than the ratio associated with board games. n330 Consequently, although holding parody unfair and board games fair would be less efficient, given current tastes, than the opposite decision, the change in popular preferences caused by the ruling would soon make adhering to it more efficient than reversing it. Moreover, the judge concludes that, when the ruling's short-term efficiency losses and long-term efficiency gains (as compared to the decision he was initially inclined to make) are reduced to present values, the latter substantially exceed the former. What should the judge do?

The foregoing example may seem fanciful, but analogous (albeit less clear-cut) problems arise frequently in fair use cases. Virtually any decision that may significantly affect the availability or price of an art form can affect popular tastes for it. A judge called upon to decide such a case must first make an effort to predict the impact on popular preferences of alternative rulings, n331 and then devise a system for choosing between those trajectories. n332

Assume, for the moment, that the judge succeeds in the first enterprise and has at least a rough idea of how the rules he is considering would affect tastes. On what basis should he make his decision? The literature on "law and economics" is surprisingly barren of reflection on that question. n333 What follows are some preliminary speculations on the issue.

[*1737] The style of economic analysis employed in these pages takes consumers' desires as given. n334 On that premise, it might seem that whichever decision in a fair use case promises to make people happier -- judged by the value they will place upon the menu of intellectual products available to them in the future -- ought to be adopted. n335 So, in the case described above, the judge should declare parody unfair and board games fair.

What that approach ignores, however, is that consumers are likely to have opinions regarding whether and how they want their tastes reshaped -- so-called "preferences about preferences" -- and that those opinions deserve at least as important a role in the economic calculus as first-order preferences. n336 Just as we respect an individual's decision to go to law school rather than business school -- even (indeed especially) if it is motivated by a desire to have his character modified in a particular way -- so ought we to defer to the desires of members of the public regarding whether they want to be sensitized to the advantages of opera, desensitized to the pleasures of parody, etc. If this analogy holds, a judge confronted with a fair use case would have to ascertain and take into account the attitudes of the persons who will be affected by his decision toward alternative ways of refashioning their tastes and the amounts they would pay to have those attitudes respected.

Much of the appeal of the foregoing solution derives from its apparent conformity with the central principle of modern liberalism [*1738] -- "that the government [should] treat all those in its charge as equals, that is, as entitled to its equal concern and respect." n337 A corollary of that principle, it is commonly thought, is that "government must be neutral on . . . the question of the good life" -- must not give preference in its decisions to any particular vision of what gives value to life. n338 Fidelity to that notion might seem to require that government, represented by the judge in a fair use case, defer to consumers' current attitudes regarding the sorts of persons they wish to become.

Further analysis, however, reveals two defects in the proposed procedure. First, its conformity with liberalism proves illusory. If the judge's decision could be expected to affect the tastes of only one person, deference to that person's second-order preferences might well be the course most respectful of his autonomy. Most decisions, however, will affect the tastes of large numbers of persons, whose second-order preferences differ. Under those circumstances, deference to the convictions of the majority (or worse yet, the minority) will give rise to the liberal nightmare -- the imposition on some persons of others' conception of the good. n339

Second, consumers' second-order preferences may prove as malleable as their primary tastes. If so, the judge would have to consult their third-order preferences to choose between alternative ways of shaping their second-order preferences. The danger of an infinite regress, at each stage of which the judge's capacity to ascertain persons' desires diminished, should be apparent.

The latter point has additional, more fundamental ramifications. If it turns out that not only our primary desires but also our views of the sorts of persons we wish to become are influenced by law, and if the legitimacy of the laws by which our outlooks have already been molded cannot be taken for granted, n340 then the reasons why the fair use doctrine (or any other doctrine) should be modified to increase the satisfaction of our extant tastes become obscure. n341 Put differently, [*1739] recognition of the ubiquity of preference-shaping casts doubt on the principle of consumer sovereignty upon which the currently dominant form of economic analysis of law rests. If the argument developed in the preceding forty pages is not to be rendered irrelevant, it is imperative that we identify some standard of value independent of consumers' current desires that can temper or set into context the wealth-maximizing principle to which we have thus far paid obeisance. n342 The job of developing such a standard is begun in Part V. Before taking up the task, however, we should pause to consider the lessons of the investigation we have just completed.

E. The Value of the Method

Few judges would be willing, when confronted with a copyright infringement case in which the defendant invoked the fair use defense, to attempt an analysis of the sort sketched in sections B and D. Of those inclined to commence such a project, few would succeed. To ascertain the economically optimal package of entitlements for a given type of copyrighted work using this mode of analysis, the judge would need an extraordinary amount of information, much of it very hard to obtain. Even determining popular tastes for different sorts of intellectual products and their uses would be difficult. Gathering reliable data with regard to such crucial variables as the sensitivity of different types of artists to fluctuations in their anticipated incomes n343 and the mutability of each of the relevant sets of preferences would probably be impossible. n344

[*1740] The fact that it would be infeasible to use the method outlined above to generate a fair use doctrine that would maximize allocative efficiency, however, does not imply that the approach is useless. On the contrary, the method points to several ways in which the doctrine could be amended that would substantially reduce the waste in resources it currently produces. The most important of those guidelines are reviewed below.

1. Arranging Entitlements in the Right Order. -- If two putatively infringing activities pose comparable dangers to the incomes of the owners of copyrights in a particular type of work, and if the incentive/loss ratio associated with the first exceeds the ratio associated with the second, a rule that the first is fair and the second unfair is less efficient than a rule that the first is unfair and the second fair. Courts and commentators often violate this relatively simple guideline because they fail to take into account all of the potential gains or efficiency losses caused by proscribing a given activity. The three most common and serious of those errors (recognition of which is facilitated by the foregoing analysis) are as follows.

(a) Narrow Conceptions of "Harm" to Copyright Owners. -- The numerator of the ratio associated with a putatively infringing use is the gain to creators of a particular type of work that would result from forbidding the use. n345 The reason we are interested in such "gains" is that, in combination, they constitute the monetary incentive for creating such works. n346 It follows that any predictable improvement in the welfare of creators caused by forbidding a use should count as a "gain." n347 Several of the hypothetical examples discussed in the preceding pages have illustrated that principle.

When defining the "injury" to copyright owners caused by declaring particular uses fair, courts often ignore this point. For example, in Consumers Union v. General Signal Corp., n348 the publisher of Consumer Reports sought an injunction against the broadcast of a vacuum-cleaner advertisement that quoted some passages from the magazine's favorable review of the product. The plaintiff argued that consumers seeing the advertisement would suspect that the manufacturer of the vacuum cleaner had paid Consumer Reports to endorse the product and would lose faith in the objectivity of the magazine's reviews of other products. The Second Circuit vacated the district [*1741] court's grant of a preliminary injunction largely on the grounds that "[t]he copyright laws . . . are not aimed at recompensing damages which may flow indirectly from copying" n349 and that the plaintiff had alleged, at most, a likelihood that "the value of possible future issues of Consumer Reports" would be impaired by the advertisement, whereas the law was concerned only with "the effect of the use upon the potential market for or value of the copyrighted work." n350 If the plaintiff's fears were justified, the loss of revenue to publishers of consumer magazines caused by legalizing the defendant's behavior could well be enormous, and the incentive/loss ratio associated with forbidding the practice could be high. n351 By refusing to recognize as "harms" the injuries predicted by the plaintiff, the court implicitly treated the ratio associated with the activity as zero, and may well have wrought considerable inefficiency by its decision. n352

Commentators are equally prone to error on this issue. Most articles that attempt to determine whether a particular activity should be deemed a fair use concentrate on the threat the activity poses to copyright owners' ability to exploit the "primary" or "ordinary" markets for their works, and deliberately or unconsciously exclude from their analysis other ways in which the activity in question could affect the welfare of the copyright owners. n353

[*1742] (b) Insensitivity to Price Discrimination and Infringers' Market Power. -- In two circumstances discussed above, the behavior of a copyright owner or putative infringer affects the economic impact of declaring an activity unlawful. The first of these is the practice of price discrimination. The ability of a copyright owner to demand more for access to his work from persons who are able and willing to pay more is desirable from an economic standpoint because it enables the copyright owner to reap greater rewards per unit of efficiency loss than he would if he charged a flat fee. n354 Accordingly, judges should watch for situations in which unauthorized use of copyrighted material undermines price discrimination schemes and should be chary of holding such uses fair.

The other circumstance is illustrated by the first and second rounds of discussion of the Barker Brothers case. n355 The upshot of those analyses was that, in situations in which a putative infringer is reaping monopoly profits in the sale of a product derived in some way from a copyrighted work, the economic impact of holding his activity unfair will depend heavily not only on whether the copyright owner and infringer would negotiate a licensing arrangement, but also on the terms of that arrangement. Generally speaking, the more sophisticated the parties, the higher the incentive/loss ratio will be. n356 Courts should thus be watchful for categories of cases in which copyright [*1743] owners have an interest in permitting infringers who enjoy significant market power in the sale of a derivative product to continue to manufacture those goods and in which all relevant parties are shrewd economic actors. In such contexts, the courts should be extremely reluctant to hold the use in question fair.

(c) Refusal to Consider the Creative Character of the Use. -- Section II.C.6 showed that, prior to 1984, many courts and commentators maintained that "productive" or "intellectually transformative" uses of copyrighted material should be deemed "fair" more readily than "nonproductive" uses. One of the effects of Justice Stevens' majority opinion in Sony has been to undermine that view. The economic analysis conducted in this Part suggests that a version of the "productivity" factor should be resurrected. As was observed in section IV.D.5, transformative uses of copyrighted material are as likely to be "public goods" as works created out of whole cloth. Holding such uses unfair reduces the rewards available to persons who engage in them, and thus in the long run may prevent the creation of intellectual products worth more to consumers than the costs of producing them. In at least a rough way, courts could and should take that consideration into account when assessing fair use claims: other things being equal, n357 an activity that improves upon or makes some creative use of copyrighted material should be permitted more readily than an activity that does not. n358

2. Rules, Categories, and Precedents. -- One of the lessons of Part II was that the Supreme Court's ratification, in Sony and Harper & Row, of the traditional "case-by-case" approach to the application of the fair use doctrine has had unfortunate effects in practice. n359 The economic analysis conducted in this Part reinforces that conclusion. Not only does ad hoc decisionmaking in this context entail substantial administrative costs, but it also prevents artists and potential artists from knowing in advance the rewards they stand to reap by devoting [*1744] their energies to creating particular kinds of products. It thus impedes the ability of the copyright system to foster the production of works of the intellect from which we all benefit.

The discussion in this Part also suggests why the Court was reluctant to depart from the traditional approach. An extraordinary variety of intellectual products are governed by the copyright system. The creators of different types of works are responsive to economic and noneconomic incentives in widely varying degrees, and the threats to their livelihoods posed by different kinds of unauthorized uses of their products vary dramatically by context. The best way to tune the copyright system, it might appear, would be to resolve controversies on an individual basis.

Last but not least, the analysis has offered some ways of resolving the tension between the points made in the preceding two paragraphs. Specifically, section IV.D.1, after identifying the economic advantages of prescribing rules that indicate which uses of copyrighted material are permissible and which are not, set forth a method for determining when to subdivide categories of copyrighted works and their uses. Section IV.D.2 then demonstrated that, once rules governing particular types of works and uses are established, they should rarely be changed.

3. Collateral Benefits. -- In addition to suggesting the foregoing ways in which fair use cases could be resolved more efficiently, the economic analysis just completed contributes in two respects to the project of this Article. First, it provides a methodology and several specific suggestions n360 that could be put to good use by Congress when it next overhauls the copyright statute. Second, it provides an analytical framework -- an understanding of the structure and function of copyright law in general and the fair use doctrine in particular -- that will stand us in good stead when considering how the law might advance a more inclusive set of values. To that project we now turn.

* Assistant Professor, Harvard Law School. For financial support in writing this Article, I am grateful to the Harvard Law School Faculty Summer Research Program and to the Harvard Program in Law and Economics, which is funded in part by the John M. Olin Foundation. Access granted by Mead Data Central to the Lexis/Nexis services facilitated the research. Preliminary versions of Parts IV and V were presented to the Real Property Section of the American Association of Law Schools in January 1987. A more developed version of Part IV was presented to the Harvard Law and Economics Seminar in April 1987. The Article has benefitted considerably from the reactions of the participants at the two meetings. In addition, I am grateful for the comments and suggestions of Robert Clark, Gerald Frug, Wendy Gordon, Benjamin Kaplan, Louis Kaplow, Reinier Kraakman, Frank Michelman, Richard Posner, Margaret Jane Radin, Eric Rakowski, Steven Shavell, Richard Stewart, Cass Sunstein, Daniel Tarullo, Lloyd Weinreb, and David Westfall. Bibliographic research and perceptive criticism by Price Marshall substantially improved the argument of Part V. The secretarial assistance of Kathy Maloney was invaluable.

n194 Calabresi & Melamed, Property Rules, Liability Rules, and Inalienability: One View of the Cathedral, 85 HARV. L. REV. 1089, 1094 (1972). Among the aliases for this conception of efficiency are the "Kaldor-Hicks criterion," the "potential Pareto superiority test," and the "wealth maximization criterion." See Bebchuk, The Pursuit of a Bigger Pie: Can Everyone Expect a Bigger Slice, 8 HOFSTRA L. REV. 671, 671 n.2 (1980); Coleman, Economics and the Law: A Critical Review of the Foundations of the Economic Approach to Law, 94 ETHICS 649, 651-52 (1984); Posner, Utilitarianism, Economics, and Legal Theory, 8 J. LEGAL STUD. 103, 119-35 (1979). For divergent treatments of the relationship between this definition of efficiency and the concepts of "Pareto superiority" and "Pareto optimality," see B. ACKERMAN, ECONOMIC FOUNDATIONS OF PROPERTY LAW xi-xiv (1975); Coleman, supra, at 649-52; Kronman, Wealth Maximization as a Normative Principle, 9 J. LEGAL STUD. 227, 229-34 (1980); Posner, The Ethical and Political Basis of the Efficiency Norm in Common Law Adjudication, 8 HOFSTRA L. REV. 487, 488-91 (1980).

n195 No effort is made in this paper to consider the impact of alternative formulations of American copyright law on the production or dissemination of works of the intellect in other countries. For some speculations on that topic, see Adelstein & Peretz, The Competition of Technologies in Markets for Ideas: Copyright and Fair Use in Evolutionary Perspective, 5 INT'L. REV. L. & ECON. 209 (1985).

n196 Following the lead of most economic analysts of the law, this article will most often use "offer" rather than "asking" prices when the two measures diverge. In other words, when the amount of money a person would offer to purchase a good or service if he were not already entitled to it differs from the amount he would demand in return for surrendering the good or service if he were entitled to it, the former figure will be used. For discussion of when and why these two figures will differ, see, for example, Kelman, Consumption Theory, Production Theory, and Ideology in the Coase Theorem, 52 S. CAL. L. REV. 669 (1979), and Spitzer & Hoffman, A Reply to Consumption Theory, Production and Ideology in the Coase Theorem, 53 S. CAL. L. REV. 1187 (1980). For a discussion of the problems attendant upon the (necessarily arbitrary) choice of one or the other measure, see Baker, The Ideology of the Economic Analysis of Law, 5 PHIL. & PUB. AFF. 3 (1975).

n197 For a general discussion of public goods and the special challenges they present to economic analysts and policymakers, see E. MANSFIELD, MICROECONOMICS 481, 489-95 (5th ed. 1985); Arrow, Economic Welfare and the Allocation of Resources for Invention, in THE RATE AND DIRECTION OF INVENTIVE ACTIVITY: ECONOMIC AND SOCIAL FACTORS 609-25 (National Bureau of Economic Research 1962); Samuelson, The Pure Theory of Public Expenditure, 36 REV. ECON. & STAT. 387 (1954).

n198 The development of increasingly convenient and inexpensive copying technologies has exacerbated this difficulty. See Liebowitz, supra note 185, at 184; Menell, Tailoring Legal Protection for Computer Software, 39 STAN. L. REV 1329, 1337-38 (1987).

n199 See B. V. HINDLEY, THE ECONOMIC THEORY OF PATENTS, COPYRIGHTS, AND REGISTERED INDUSTRIAL DESIGNS 1-33 (Economic Council of Canada 1971); Gordon, supra note 185, at 1610-12.

n200 If the creator is in the business of manufacturing and selling physical embodiments of his work (e.g., the publication of copies of a book or the production of "floppy disks" containing a software program), this "marginal cost" will be a positive (and usually relatively stable) number. If the creator is merely granting permission to use his work (e.g., patent licensing or authorizing the copying of software programs), "marginal cost" will be zero. Most of the graphs and illustrations in this section treat marginal cost as positive and constant. None of the conclusions would change if marginal cost were zero or variable.

n201 This strategy may be represented graphically as follows:

The level of output that will enable the monopolist to maximize his profits is indicated by point E; the corresponding monopoly price is indicated by point A. For more elaborate discussion of profit maximization by a monopolist, see J. HIRSHLEIFER, PRICE THEORY AND APPLICATIONS 238-42 (3d ed. 1984).

The marginal revenue (MR) curve in Figure 1 declines more steeply than the demand (D) curve because, in order to sell an additional copy of the work, the copyright or patent holder ordinarily must lower the price charged to all purchasers of the work. To the extent that the copyright or patent holder is able to engage costlessly in price discrimination (adjusting the price he charges each consumer to match the value the consumer places on the work), the foregoing generalization does not hold, the marginal revenue curve will approach the demand curve, and the "deadweight loss," see infra p. 1702, ordinarily associated with the exercise of monopoly power will diminish. See J. HIRSHLEIFER, supra, at 255-61. The implications for the design of the fair use doctrine of the availability of opportunities for price discrimination are discussed at pp. 1709-10 below.

n202 These so-called "monopoly profits" are represented by the rectangle ABDC in Figure 1.

n203 In other words, they will either go without intellectual products of that sort altogether, or they will shift to what they regard as less satisfactory substitutes.

n204 See, e.g., R. POSNER, ECONOMIC ANALYSIS OF LAW 256 (3d ed. 1986) (treating "the transfer of wealth from consumers to producers brought about by increasing the price from the competitive to the monopoly level . . . as a wash" for the purposes of the "economic . . . conception[] of welfare"). As Judge Posner acknowledges, see id., at 257-59; Posner, The Social Costs of Monopoly and Regulation, 83 J. POL. ECON. 807, 807-15 (1975), in some contexts the existence of these monopoly profits may lead to efficiency costs -- for example, by reducing the monopolist's incentive to innovate. The magnitude and even the existence of these consequential losses are disputed, however, and the debate is especially inconclusive as regards patent or copyright monopolies. See F. M. SCHERER, INDUSTRIAL MARKET STRUCTURE AND ECONOMIC PERFORMANCE 450-54 (2d ed. 1980).

The fact that, in the view of most economists, this transfer of wealth from consumers to producers by itself has no impact on efficiency does not mean that it need not be taken into account in an economic analysis of intellectual property law; on the contrary, it is the source of the monetary incentive emphasized in the preceding paragraph. Thus, much of the discussion in this Part of the Article will concern the relationships in various contexts between the magnitude of this transfer and the magnitude of the concomitant monopoly losses.

n205 In Figure 1, the lost consumer surplus is represented by the triangle CDF. Because, by hypothesis, marginal cost is constant, see supra note 200, the copyright holder whose activities are depicted in the figure forgoes no producer surplus by producing quantity E rather than quantity G. If marginal cost were not constant, the monopoly pricing strategy would result in a loss of producer surplus. For discussion of these effects, see Landes & Posner, Market Power in Antitrust Cases, 94 HARV. L. REV. 937, 954, 991-96 (1981).

n206 Patent law differs from copyright law in several respects. For example, patent law has stricter standards of originality and creativity, compare 35 U.S.C. §§ 102(a), 103 (1982 & Supp. IV 1986) and Graham v. John Deere Co., 383 U.S. 1, 12-19 (1966) with I M. NIMMER, NIMMER ON COPYRIGHT §§ 1.06[A], at 1-37, 1.08[C][1], at 1-48, 2.01, at 2-5 (1987), protects ideas as well as the expression thereof, compare 35 U.S.C. § 101 (1982) with Sheldon v. Metro-Goldwyn Pictures Corp., 81 F.2d 49 (2d Cir.), cert. denied, 298 U.S. 669 (1936), and excludes the second of two persons to develop an idea independently, compare 35 U.S.C. § 102(g) (1982) with 1 M. NIMMER, NIMMER ON COPYRIGHT § 2.01[A] (1987). These differences make it somewhat more likely that a patent holder will enjoy significant market power. However, the fact that many patented products and processes have what consumers regard as close substitutes means that even patent holders are not assured of such power. See Kitch, Patents: Monopolies or Property Rights?, in THE ECONOMICS OF PATENTS AND COPYRIGHTS, supra note 185, at 31-49.

n207 Examples that have figured in important fair use cases are the letters and papers of J. D. Salinger and George Washington, see Salinger v. Random House, Inc., 811 F.2d 90 (2d Cir. 1987); Folsom v. Marsh, 9 F. Cas. 342 (C.C.D. Mass. 1841) (No. 4901), the Zapruder photographs of the assassination of President Kennedy, see Time, Inc. v. Bernard Geis Assocs., 293 F. Supp. 130 (S.D.N.Y. 1968), and the movie "Gas Light," see Benny v. Loew's Inc., 239 F.2d 532 (9th Cir. 1956), aff'd per curiam by an equally divided Court sub nom. CBS, Inc. v. Loew's Inc., 356 U.S. 43 (1958). The "consumer" in the last-mentioned case who seems to have regarded the movie as virtually unique was a parodist. Moviegoers undoubtedly would have been much more likely to shift to other movies if the price of admission to "Gas Light" had risen sufficiently. The general point is that the degree of market power enjoyed by a given copyright holder may vary dramatically between different types of consumers (that is, within different markets).

n208 Examples include B-grade pornographic movies and most law-review casenotes. The holders of the copyrights in such works face nearly flat demand curves. In other words, if they raise their prices even a modest amount, the quantity of copies purchased will decline sharply because consumers will shift to works they regard as almost as good.

n209 Much of the confusion in economic analyses of the copyright system derives from the failure of most analysts to acknowledge the existence of this spectrum. A few economists take the position that there exist nearly perfect substitutes for most copyrighted works and that, therefore, a policymaker in this field need not worry about the inefficiency that would ensue if copyright holders enjoyed monopoly power. See, e.g., Smith, Collective Administration of Copyright, in THE ECONOMICS OF PATENTS AND COPYRIGHTS, supra note 185, at 139, 144. Most take the opposite position and assume that all copyright holders enjoy substantial market power. See, e.g., Liebowitz, supra note 185, at 184-86. In reality, the market power enjoyed by the holders of the copyrights in works of different sorts varies enormously, and a lawmaker who wishes to tune the copyright system to maximize efficiency must take those variations into account. See infra pp. 1720-21.

n210 For economic analyses of intellectual property regimes that describe the lawmaker's task in similar terms, see Kaplow, The Patent-Antitrust Intersection: A Reappraisal, 97 HARV. L. REV. 1813, 1827-28 (1984); Liebowitz, supra note 185, at 184-88; Menell, supra note 198, at 1367-69.

n211 See supra note 205 and accompanying text.

n212 For example, he could adjust the duration of copyrights (in general or in particular sorts of works); he could permit some uses of copyrighted works (e.g., public performance and display) while forbidding others (e.g., copying, preparing derivative works, or borrowing a copy from a public library without paying a fee); and he could regulate the prices a copyright owner could charge for permission to engage in particular uses of his work.

n213 For example, the options discussed in note 212 above are settled by sections 302, 106, 111(d), 115, 116, and 118 of the current statute. In contrast to the copyright statutes in several European countries, American copyright law does not accord a copyright owner "public lending rights," enabling him to charge persons who borrow his works from libraries. See Treece, Library Photocopying, 24 UCLA L. REV. 1025, 1025-26 (1977).

n214 For the text of the provision, see note 4 above. Each of the factors identifies an area of inquiry but does not purport to determine its relevance. In addition, the relative weights that the four factors should be accorded in the analysis are unspecified.

n215 See supra sections II.B.3 and II.B.5 (discussing factors the Supreme Court has already added to the list contained in the statute).

n216 See H.R. REP. NO. 1476, supra note 9; S. REP. NO. 473, supra note 9.

n217 See 17 U.S.C. § 107 (1982).

n218 Cf. Liebowitz, supra note 185, at 188 (proposing a similar interpretation of the fair use doctrine). But cf. Gordon, supra note 185, at 1613 n.81 (arguing that courts interpreting the fair use doctrine should not strive to achieve "optimal" levels of copyright protection, but rather should accept that "the copyright law treats the outcome of the ordinary copyright transaction as normatively equivalent to an 'optimal' result," and should recognize the fair use defense only in "settings" different from "an ordinary transactional setting"). Justice Kaplan has suggested that we "dispense" with the fair use doctrine altogether and consider the "policy" questions "fair use" has come to incorporate under the more general doctrinal heading of "infringement." See B. KAPLAN, supra note 144, at 67-70. This and the succeeding Parts of the Article consider an alternative way of partially combining these two strands of copyright law -- namely, by subsuming, under the heading of "fair use," many of the questions currently addressed by courts when determining whether a copyright has been "infringed."

n219 See 17 U.S.C. § 504(a) -- (c) (1982).

n220 "Fair use is a mixed question of law and fact." Harper & Row, 471 U.S. at 560. Either party, consequently, may insist that the issue be decided by a jury. See A. LATMAN, THE COPYRIGHT LAW 295-97 (W. Patry 6th ed. 1986); W. PATRY, supra note 6, at 480. For simplicity, it will be assumed throughout this and subsequent Parts that (as is typical in copyright cases, see A. LATMAN, supra, at 295) all parties waive their rights to a jury trial and that the judge must therefore both ascertain the facts of the cases and determine their legal significance.

n221 Cf. A. POLINSKY, AN INTRODUCTION TO LAW AND ECONOMICS 2-4 (1983) (describing the use of assumptions in economic analysis); Spitzer & Hoffman, supra note 196, at 1190-91 (recommending proceeding in stages of this sort).

n222 See infra section IV.C.

n223 The reader may wonder why the relevant category is "detective stories." That issue is taken up in section IV.D.1 below.

n224 Taking into account (i) that the Copyright Act accords a copyright owner the exclusive right "to prepare derivative works based upon the copyrighted work," see supra p. 1662; (ii) that many of the components of a board game (albeit not the game itself) are copyrightable, see 1 M. NIMMER, NIMMER ON COPYRIGHT § 2.18[H][3][a], at 2-213 (1987); and (iii) that substantial mimicry of the plot of a copyrighted work will be deemed infringement, see, e.g., Sheldon v. Metro-Goldwyn Pictures Corp., 81 F.2d 49 (2d Cir.), cert. denied, 298 U.S. 669 (1936), Barker's conduct most likely would be actionable were it not excused as a fair use.

n225 Why the judge should consider a writer of Plaintiff's stature when making these calculations is taken up at p. 1715 below.

n226 Shouldn't the judge be interested in the efficiency gains from holding out that reward, not in the reward itself? The answer is yes, see supra p. 1703, but those efficiency gains cannot be determined without knowing how much money writers could make if they could not demand fees from this particular group of users. (The increase in the production of detective stories caused by raising the average annual income of writers from $10,000 to $20,000 is likely to be much greater than the increase caused by raising their average annual income from $100,000 to $110,000.) That depends, in turn, on what other putatively infringing activities are declared fair and unfair -- which is precisely what the judge does not yet know. Only after the judge has determined the impact upon authors' incomes of permitting or forbidding each activity will he be in a position to ascertain the optimal package of entitlements. See infra pp. 1715-16.

n227 This approach parallels in several (but, as will be seen, not all) respects the method developed by Professor Kaplow in his analysis of patent law, see Kaplow, supra note 210, at 1829-34 & n.54.

n228 It is customary for an author's royalties to be set at a percentage of the retail price of the book. See Brockway, Business Management and Accounting, in WHAT HAPPENS IN BOOK PUBLISHING 200-02 (C. Grannis ed. 1967). The alternative arrangement described in the text -- in which Plaintiff pays her publisher a fixed amount for each book sold and keeps the remainder of the revenues -- is selected in order to simplify the discussion and to facilitate illustration of the central economic principles. For the same reasons, it is also assumed that Plaintiff is not obliged to pay her publisher anything for books printed but not sold. For discussion of the divergence of interest between author and publisher under the typical royalty arrangement, see J. HIRSHLEIFER, cited above in note 201, at 247-48.

n229 The possibility that Plaintiff anticipated releasing the book later in another form is considered at pp. 1718-22 below.

n230 It is assumed, for the sake of simplicity, that both Plaintiff and Pirate have complete control over the retail prices of their editions, that the production costs of Plaintiff's publisher and Pirate are similar, and that the profit margins demanded by distributors and retailers for carrying the two editions are the same.

n231 Professors Hurt and Schuchman and Judge Breyer argue that the ability of publishers to engage in price competition of this sort and in other forms of "retributive" behavior would deter copyists like Pirate from producing unauthorized, lower-priced editions even if the copyists were not restrained by the copyright laws. See Breyer, supra note 144, at 300-01; Hurt & Schuchman, supra note 185, at 428. For a convincing rebuttal, see Tyerman, cited above in note 148, at 1112-13.

n232 Assume the judge's research exposes the following economic conditions:

Although some mystery-story readers have already purchased copies of More Adventures, many have not. The demand curve in Figure 2 represents the remaining demand for the book. The marginal cost curve (which represents the cost, from Plaintiff's standpoint, of producing and selling each additional copy) is derived from the provision of the publishing contract governing the portion of the purchase price retained by the publisher, see supra p. 1707, plus the portion of the price retained by the distributors and retailers, see supra note 230.

n233 Plaintiff's monopoly profits are represented by the area of rectangle ABDC. The associated deadweight losses are represented by area of triangle CDF. On the assumptions embodied in Figure 2, the former is twice the size of the latter. Were the demand curve shaped differently or were the marginal cost curve shaped or sloped differently, that figure could be significantly different. Here, as in the rest of Part IV, what is important is not the numerical ratios suggested by the illustrations, but the method of deriving them.

n234 More specifically, she is engaged in what is known as "third-degree price discrimination" partly through "product differentiation." See E. MANSFIELD, supra note 197, at 296-99; F. M. SCHERER, supra note 204, at 316-19.

n235 In the eyes of some consumers, hardcover editions have sufficient advantages over paperback editions because of durability, larger print, or snob appeal to be worth the extra money. See Tyerman, supra note 148, at 1110. However, many if not most purchasers of hardback editions of popular tradebooks would purchase paperback editions if they were available.

n236 That price discrimination increases monopoly profits is indisputable. See, e.g., E. MANSFIELD, supra note 197, at 297; F. M. SCHERER, supra note 204, at 319-20. The direction and magnitude of the associated effect on deadweight losses is not so clear; economists differ as to whether, assuming a manufacturer continues to enjoy monopoly power, allocative efficiency is enhanced by allowing or forbidding him to engage in third-degree price discrimination. See F. M. SCHERER, supra note 204, at 320-22 (discussing the literature). However, the former effect is so substantial that, even if price discrimination by Plaintiff increased the misallocation of resources to a modest degree, the incentive/loss ratio would almost certainly rise. Cf. Kaplow, supra note 210, at 1875 (making a similar argument regarding the desirability of permitting patentees to engage in price discrimination).

n237 If Plaintiff plans to sacrifice to Bestseller the portion of the market for her book that consists of Bestseller's members, why would Plaintiff release her own paperback edition any earlier than she had originally planned? The answer (the judge discovers) is that, if the only two versions of the book available to consumers are Plaintiff's hardcover edition and Bestseller's much cheaper paperback edition, Bestseller's share of the market will expand in two ways: more readers will join Bestseller Book Club, and retail bookstores will contrive ways of obtaining copies of Bestseller's paperback edition. It is those two effects that Plaintiff wishes to prevent.

n238 The judge learns that $10 is the retail price at which Barker expects to maximize its profits. To preserve its position in the market, Barker might be obliged periodically to raze the undergrowth of competitors by sharply lowering its prices. See Areeda & Turner, Predatory Pricing and Related Practices Under Section 2 of the Sherman Act, 88 HARV. L. REV. 697, 697-99 (1975). However, Barker does not anticipate pursuing that strategy in its marketing of "Detection."

n239 Because Plaintiff is a monopolist and Barker is, for practical purposes, a monopsonist, it would be hard for the judge to predict the amount of the royalty they would agree upon. See E. MANSFIELD, supra note 197, at 295-96 (discussing the indeterminacy of the outcome of negotiations in situations of bilateral monopoly). Assume that the judge is somehow able accurately to estimate the royalty at $2 per game.

Support for the prediction that the parties would employ a simple arrangement of the sort described in the text may be found in Schroeder, Licensing of Rights to Intellectual Property, 50 ALB. L. REV. 455, 457-58, 460 (1986). More complex licensing schemes and their effects on the incentive/loss ratio are considered in section IV.D.4 below.

n240 This response might be depicted graphically as follows:

The price of the game would rise from point B to point A.

n241 The decline would equal the difference between the areas of rectangles BEQM and ACIG.

n242 The increase would equal the area of Figure GLRM.

n243 That profit is represented by the area of Figure DFLJ.

n244 If either the quality of games of this sort or the likelihood that they would be developed varied with the amount of profit Barker stood to reap from their sale, the judge would have to take into account the efficiency losses associated with the decline in Barker's monopoly profits. See supra p. 1711. For the time being, it is assumed that the size of Barker's income has no such effect. This assumption is removed in section IV.D.[5] below.

n245 See supra note 240.

n246 See Gordon, supra note 185, at 1634.

n247 The size of the denominator of that fraction might suggest to the judge that the comparison made in the text is misleading. If so many consumers would benefit so substantially from publication of the review, perhaps they can be charged for access to it --charged enough by Plaintiff, directly or indirectly, to offset the reduction in the demand for her book. In other words, the outcome of holding that the review was an infringement might not be its suppression, but rather the negotiation of an agreement that would satisfy (for a price) consumers' thirst for information and increase the net reward to Plaintiff. Further investigation by the judge, however, reveals that this happy scenario is unlikely to occur. Accorded the right to grant or refuse permission to print reviews, authors would be likely to withhold their consent whenever the reviewer's comments were hostile. Not only would readers thereby be denied the benefits of the information contained in negative reviews, but their awareness of the situation would impair their confidence in the accuracy of the reviews that did pass the author's muster. Thus, assuming for the moment that the judge's only options are to declare the use fair or unfair, but cf. infra section IV.D.3 (discussing mandatory licensing), the ratio stated in the text is the right one.

n248 The judge concludes that transaction costs preclude a fourth possible response: negotiation of an agreement with Plaintiff under which Plaintiff would receive a royalty for each story distributed to a student. Cf. Liebowitz, supra note 185, at 189 (suggesting that transaction costs would make it uneconomical for copyright owners to extract fees from persons who wish to photocopy their works in libraries). But cf. id. at 192-94 (sketching a price discrimination scheme that might enable copyright owners to collect such fees indirectly). The judge also concludes that a fifth possible response -- requiring students to purchase hardcover copies of More Adventures in order to read one of the stories -- would be so rare as not to affect the calculus.

n249 The reason: the substitution and exposure effects discussed above, would remain unchanged because students would still be assigned one of Plaintiff's stories, but Plaintiff's royalties from the sale of anthologies would increase somewhat.

n250 The measure of the resulting efficiency loss would be the difference between the cost of a course using photocopied materials and the price at which each deterred student valued the course.

n251 Indeed, the economic costs might be even higher; a significant number of students would now remain ignorant that they could improve their welfare by reading Plaintiff's work, which would result in an efficiency loss similar to that discussed in connection with Critic's review. See supra p. 1712.

n252 The other sources of Plaintiff's income would not be materially affected. The exposure effect described above would not charge at all. The substitution effect would increase slightly -- insofar as students enthralled by Plaintiff's writing would now be somewhat more inclined to continue reading the library copy of More Adventures and not to purchase copies of their own -- but not enough to affect the calculus.

n253 That is, those who would now be obliged to go to the library.

n254 That is, those who would have read Plaintiff's story (in photocopy form) in their rooms but who would not read the story if forced to make the trip to the library.

n255 Or, more precisely, by the percentage of the total number of students taking short-story courses who are enrolled in those professor's courses.

n256 See supra p. 1703.

n257 What this way of describing the judge's task takes for granted is that, to the extent actual and potential artists of a particular type are influenced by the prospect of monetary gain, they are affected most by the rewards available to successful or prominent artists of the sort they are considering becoming. That assumption, although plausible, is by no means obviously correct. Self-confidence, fear of mediocrity, and the "vividness" heuristic, see R. NISBETT & L. ROSS, HUMAN INFERENCE: STRATEGIES AND SHORTCOMINGS OF SOCIAL JUDGMENT 43-62, 122-27 (1980), would all lead artists to concentrate on the most accomplished members of their fields, but risk-aversion, "success anxiety," and the "representativeness heuristic," see id. at 24-28, might lead them to concentrate on more ordinary members. The judge would therefore have to ascertain through empirical investigation whether prospective detective-story writers are more concerned with the incomes of writers like Plaintiff or with those of less famous practitioners of the trade. If the latter is the case, he would have to recalculate all of the fractions discussed above, using demand functions of the sort that confront "average" detective-story writers, which would be more elastic than those confronting Plaintiff. See supra pp. 1702-03.

n258 In attempting to identify and weigh such noneconomic motives, the judge would do well to start with the list provided by Hurt & Schuchman, cited above in note 185, at 425-26:

[Such] desires include the propagation of partisan ideas; notions of altruism, as in the case of religious and moral tracts; desire for recognition; and enhancement of one's reputation. There are also cases where authors pay for the costs of publication in order to promote some other pecuniary interest than the sale of the writing itself, such as advertising copy and scholarly publication induced by the 'publish-or-perish' rule of most universities.

n259 Strickly speaking, the judge ought also to determine (and subtract from those gains) the present value of the amounts of money consumers would have been willing to pay for products or services that the mystery writers would have produced had they not been enticed into writing detective stories. See id. at 425, 429, 430. If talent were to some degree fungible -- for example, if a good mystery writer would be more likely than the average person to be a good physicist -- those costs might be significant.

n260  The scale of the X-axis was selected in order to make the Aggregate Reward function linear. Using a less arbitrary scale would not alter the location of the optimal point, but would make it more difficult to see the relationships between the different components of the judge's calculations. The utility of depicting in this fashion the choices facing the judge was suggested by a graph in Liebowitz, cited above in note 185, at 187.

n261 Although sufficient as applied to the facts as depicted in Figure 4, this procedure would have to be modified in one respect to accommodate more complex cases. The judge's objective, as the text indicates, is to select the package of entitlements whose net efficiency gain (as compared to a situation of no legal protection for intellectual property) is greatest. The procedure outlined in this section -- and, specifically, identification of the apogee of the net-economic-efficiency curve -- is a reasonably reliable way of ascertaining that package. However, if the entitlements in question are not divisible, but cf. infra section IV.D.3 (suggesting one way in which they might be divided), the proposed approach will sometimes point toward a slightly less than optimal solution.

To illustrate, assume that works of a particular sort may be infringed in four ways, and that the incentive/loss ratios associated with them are as follows: A: 5/4; B: 10/1; C: 12/2; D: 1/1. The conscientious judge arranges them in order of their ratios (B, C, A, D), constructs a graph analogous to Figure 4, determines that the net-efficiency curve peaks at use C, and is therefore inclined to declare uses B and C unfair and uses A and D fair. The differences in the magnitude of the economic impacts associated with uses A and D, however, create a possibility that reversing their order and declaring uses B, C, and D unfair and use A fair would yield greater net efficiency gains. The general lesson: the procedure outlined in this section will provide a good approximation of the most efficient solution, but a judge will sometimes be able to do a little better by altering the order of the entitlements in the near vicinity of the line between fair and unfair uses.

n262 For an argument that this may well be the case for the most popular "tradebooks" (of which More Adventures is an example), see Breyer, cited above in note 144, at 293-313. For criticism of Breyer's argument, see Tyerman, cited above in note 148, at 1108-25. It has been suggested that legal protection for most kinds of books may be inefficient for these and other reasons. See Plant, The Economic Aspects of Copyright in Books, 1 Economica 167 (1934).

n263 This seems to be the position of commentators who argue, on economic grounds, that no use of a copyrighted work should be considered fair if it results in any "significant" or "substantial" reduction in the rewards that the copyright owner could reap. See, e.g., Gordon, supra note 185, at 1618 (asserting that "[f]air use should be denied whenever a substantial injury appears that will impair incentives"); Smith, supra note 209, at 142.

n264 Although breached with notorious regularity, the principle that a court should not consider issues inessential to the resolution of the matter before it remains deeply rooted in American law. See Edwards, The Role of a Judge in Modern Society: Some Reflections on Current Practice in Federal Appellate Adjudication, 32 CLEV. ST. L. REV. 385, 410-12 (1983-84). But cf. id. at 413-20 (advocating use of a more inclusive approach, denoted "wide-angle adjudication," in special contexts.) The kind of comparative analysis outlined in section B, while not inconsistent with the letter of that principle, would carry the court further from the facts of the case at bar than many federal judges would feel comfortable going.

n265 One significant policy implication of the analysis conducted thus far will not be affected by the subsequent discussion and, because it does not bear directly on the content of the fair use doctrine, is perhaps worth mentioning at this point. Economists and legislators have long debated the optimal duration of intellectual property rights. This argument has been most heated with regard to patent life, see Kaplow, supra note 210, at 1822-29 & n.21, but has also arisen occasionally in the context of copyright law, see, e.g., Breyer, supra note 144, at 323-29. The analysis presented in section B suggests a solution to the controversy. Assume that a legislator is confronted simultaneously with two questions: (i) Which of four possible entitlements should be allocated to creators of a certain kind of intellectual product? (ii) For how long should creators enjoy those rights? The legislator begins by arranging the four entitlements in a series in order of their incentive/loss ratios (A, B, C, D). He then determines how potential creators of works of the sort in question would respond to different levels of anticipated reward. Finally, he determines what the optimal duration would be for each possible package of entitlements. His conclusions are as follows. (i) If creators were given all four rights, the optimal life would be I year. (ii) If creators were given only rights A, B, and C, the optimal life would be 2 years. (iii) If creators were given only rights A and B, the optimal life would be 10 years. (iv) If creators were given only right A, the optimal life would be infinity. The legislator clearly should select option (iv), because the high incentive/loss ratio associated with right A means that the legislator can provide greater incentives to creators at a lower efficiency cost using the fourth option than using any of the other options. The general insight that emerges from this hypothetical case is that, unless the ratios associated with different entitlements are likely to change in different ways over time, it always will be more efficient to select a long patent or copyright life and then to use the procedure sketched in section B to determine which small set of entitlements would be optimal than to select a shorter life and then to use the procedure to identify a larger optimal set of entitlements. But cf. Kaplow, supra note 210, at 1839-41 (discussing in a somewhat different manner the "Simultaneous Determination of Patent-Antitrust Doctrine and the Patent Life").

n266 This argument assumes, realistically, that potential authors are risk averse. Thus, as their future earning power becomes less certain, fewer potential authors will write, even though the expected value of their earning power remains constant.

n267 For example, writers of musicals might love their work less than writers of other plays and be more inclined to shift to other occupations if their incomes dropped a certain amount, or the profitability of musicals might be more vulnerable to impairment by an effective parody than the profitability of other plays.

n268 It is possible treating musicals and other plays differently for fair use purposes would reduce rather than increase the disparity of income that the two types of playwright could anticipate. The smaller the gap between the expected rewards associated with each category, the smaller the efficiency loss caused by writers' uncertainty as to how their creations would be classified.

n269 Several courts have made precisely this distinction. See, e.g., Walt Disney Prods. v. Air Pirates, 581 F.2d 751, 757-58 (9th Cir. 1978), cert. denied, 439 U.S. 1132 (1979); Columbia Pictures Corp. v. National Broadcasting Co., 137 F. Supp. 348, 350 (S.D. Cal. 1955); cf. Light, supra note 185, at 628-29 (criticizing the "recall or conjure up test").

n270 From an economic standpoint, it is important that the judge reveal in his opinion more than whether the defendant's conduct should be considered fair or unfair. Assume that, after conducting a comprehensive analysis, a judge concludes that copyrights in works of the sort at issue in the case before him might be infringed in 5 ways -- A, B, C, D, and E (in order of descending ratios); that the defendant has engaged in use D; and that the optimal combination of unfair uses is A + B + C. Consequently, the judge rules in favor of the defendant. At a minimum, it is imperative that the judge disclose in his opinion his conclusion that uses A, B, and C should be deemed unfair, because a decision in a subsequent case that any of those uses is fair might well render inefficient the rule that use D is fair.

n271 Should a judge's respect for precedent vary depending on whether the rule he considers abandoning is a "holding" (as would be a ruling that educational copying is fair) or "dicta" (as would be a ruling that making a derivative game is unfair)? From an economic standpoint, there would seem to be no justification for such a distinction, insofar as the ranking of each use was essential to the ultimate holding that the use in question was either fair or unfair. The only conceivable reason for differentiating the two types of ruling is that the artists themselves would consider "dicta" less permanent and therefore would not assume that departure from "dicta" implies that the "holdings" are unstable.

n272 A practical implication of this insight is that, if a regime of the sort sketched in this Part of the Article were put into place, litigants would rarely attempt to challenge an established rule governing unauthorized uses of a particular kind of copyrighted work because their chances of success would be so small. Thus, the typical fair use case would not become "a complex economic battleground populated by expert witnesses expounding philosophies allegedly supporting or rejecting recovery," as some critics of economic analysis in this context have charged. See W. PATRY, supra note 6, at 456 n.520. To be sure, fair use cases in fields not yet governed by a rule would be enormously complex affairs. See infra p. 1739. Once a rule was adopted, however, most subsequent cases in the area would be comparatively simple.

n273 This nomenclature is derived from the famous article by Calabresi and Melamed, cited above in note 194, where the phrases are defined as follows: "An entitlement is protected by a property rule to the extent that someone who wishes to remove the entitlement from its holder must buy it from him in a voluntary transaction in which the value of the entitlement is agreed upon by the seller. . . . Whenever someone may destroy the initial entitlement if he is willing to pay an objectively determined value for it, an entitlement is protected by a liability rule." Id. at 1092. For other discussions of the issues considered in this subsection, see 3 M. NIMMER, NIMMER ON COPYRIGHT § 13.05[E][4][e], at 13-111 (1987); Gordon, cited above in note 185, at 1622-24; Timberg, A Modernized Fair Use Code for the Electronic as Well as the Gutenberg Age, 75 NW. U.L. REV. 193, 233-44 (1980).

n274 To illustrate, the market for the right to make copies of a computer chess program that was sufficiently better than its competitors to accord its owner substantial market power might be represented graphically as follows:

Assuming (for the sake of simplicity) that no users will engage in unauthorized copying and that, therefore, the marginal cost to the copyright owner of authorizing copying is zero, the rational owner will grant permission only up to the point where he reaps no marginal revenue from an additional grant (quantity E, price A). The ratio of reward to monopoly loss associated with that pricing strategy is the ratio of the area of rectangle AOEC to the area of triangle CEH, which on the arbitrary assumptions embodied in the graph would be 2:1. If, by employing a liability rule, a judge forces the copyright owner to "sell" the right to make copies of the program for a price of B, the copyright owner will have no choice but to grant permission to G users, and the ratio of reward to monopoly loss will be the ratio of the area of rectangle BOGF to the area of triangle FGH -- a much larger figure.

n275 In the figure above, the area of rectangle BOGF is less than the area of rectangle AOEC.

n276 For discussion of the economic losses associated with this impact, see p. 1713 above.

n277 It is important to keep in mind that the relevant figure is the net reward reaped by the copyright owner. Thus, determination of the numerator of the fraction would require subtracting from the owner's proceeds the cost to him of monitoring the activities of users, asserting his rights, and collecting payments.

n278 The losses generated by such a rule would include the administrative and litigation costs associated with its enforcement, including those incurred by the author, see supra note 277. Those costs would probably be larger than the costs generated by a property rule, and thus would reduce the potential gains of a shift to a compulsory licensing system.

n279 If the economic consequences of forcing Plaintiff to grant Defendant permission to copy her stories in return for a fee of fifty cents a copy are so superior to those associated with denying permission outright, why would the parties not reach an agreement to that effect without the court's aid? Put differently, what are the "transaction costs" preventing such a bargain, see supra note 248, that the court is able to overcome with the compulsory license? There are two possibilities. First, Plaintiff may be inclined to engage in strategic behavior -- for example, to demand a high fee in order not only to maximize her revenues in this exchange, but also to set a "precedent" for her dealings with other teachers in the future. See generally Cooter, The Cost of Coase, 11 J. LEGAL STUD. 1 (1982). Second, Plaintiff's net income under the compulsory licensing scheme may be smaller than the injuries she sustains as a result of permitting teachers to copy her stories. (For a summary of those injuries, see the text following note 247 above.) Thus, Plaintiff, if protected by a property rule, would not voluntarily agree to a regime under which she received only fifty cents per copy.

n280 See, e.g., 17 U.S.C. §§ 111(d), 115, 116, 118 (1982 & Supp. IV 1986) (governing, respectively, cable television, phonorecords, jukeboxes, and noncommercial broadcasting). The royalty rates under these various provisions are periodically reviewed and adjusted by the Copyright Royalty Tribunal. See 17 U.S.C. § 801 (1982 & Supp. IV 1986). For an overview of these systems, see generally R. BROWN & R. DENICOLA, CASES ON COPYRIGHT 407-25 (4th ed. 1985); Greenman & Deutsch, The Copyright Royalty Tribunal and the Statutory Mechanical Royalty: History and Prospect, 1 CARDOZO ARTS & ENT. L.J. 1 (1982).

n281 The economic advantages of a compulsory licensing system of this sort would be enhanced if Congress, when establishing an administrative tribunal responsible for regulating royalty rates, staffed it with persons capable of undertaking calculations of the sort discussed in the text and provided it with better guidance regarding how prices should be set. In the pertinent provisions of the 1976 Act, Congress gave the Copyright Royalty Tribunal (and the courts that review the decisions of the Tribunal) virtually no instruction on this score. See National Cable Television Ass'n v. Copyright Royalty Tribunal, 724 F.2d 176, 182, 190 (D.C. Cir. 1983) (cable television); Recording Indus. Ass'n v. Copyright Royalty Tribunal, 662 F.2d 1, 4-5, 8-9 (D.C. Cir. 1981) (phonorecords). The Tribunal consequently has felt free to set the license fees at levels it believes an "unregulated" market would produce. See 47 Fed. Reg. 52,146 (1982) (subsequently codified as amended at 37 C.F.R. § 308 (1987)) (announcing an intention to set on this basis the rates paid to copyright owners by cable television stations). For the reasons suggested above, see supra note 274 and accompanying text, the incentive/loss ratios associated with these activities could be dramatically improved if the rates were set below the levels that copyright owners, enjoying significant market power, and persons desiring access to their works would agree upon.

n282 The plaintiffs in the Sony case argued that the district court had such a power with regard to copying of copyrighted material broadcast over the airwaves. Although the Ninth Circuit insisted that the availability of such a remedy was not essential to its result, it appeared to agree with the plaintiffs. See Universal City Studios, Inc. v. Sony Corp. of America, 659 F.2d 963, 976 (9th Cir. 1981), rev'd, 469 U.S. 417 (1983). Although commentators continue to dispute the question, those who contend that a court has the power to deny an injunction and instead order the payment of a royalty, see Nimmer, Copyright Liability for Audio Home Recording: Dispelling the Betamax Myth, 68 VA. L. REV. 1505, 1530-32 (1982), have the better of the argument. But cf. Note, The Betamax Case, supra note 13, at 261 (arguing that a court lacks statutory authority to set up a compulsory royalty scheme).

n283 Cf. Gordon, supra note 185, at 1623-24 (arguing on the basis of institutional competence that such an approach is problematic).

n284 Professor Nimmer, in discussing the desirability of such a court-ordered remedy in the context of audio home recording, asserts confidently: "Parties to subsequent actions of a similar nature could agree to be bound by the same consent decree."nimmer, supra note 282, at 1534. The author does not explain, however, why such parties would agree to abide by the terms of the original order.

n285 See id. at 1529-30.

n286 See supra pp. 1710-12.

n287 See supra note 248.

n288 See supra p. 1710.

n289 Participation in the bargaining process by lawyers will also reduce the probability that the parties, each of whom is trying not only to capture the entire value of the licensing arrangement, but also to conceal from the other the terms to which he would consent, will fail to reach an agreement. See Cooter, supra note 279, at 22.

n290 The reason that Plaintiff and Barker will seek to prevent the retail price from rising is that they both have an interest in maximizing the size of the pie to be divided. By hypothesis, $10 is the price at which the total profits from the sale of "Detection" will be greatest. See supra note 238. Both parties therefore have an interest in keeping the price at that level.

n291 There are various ways the parties could achieve these results; the simplest is an agreement under which Barker promises to pay Plaintiff a fixed annual fee and not to raise the retail price.

n292 See supra note 239.

n293 If the judge had previously concluded that Barker's activity should be declared unfair, see supra p. 1717, this change in the order of the uses would have no effect on the judge's decision. By contrast, had the apogee of the net-efficiency curve been to the left of Barker's activity, the change would probably lead to a different outcome.

n294 Organizations of both types -- known respectively as ASCAP and BMI -- have arisen in the context of the use of copyrighted musical compositions by television and radio stations, nightclubs, and restaurants. For discussion of the histories and current structures of the two organizations, see R. BROWN & R. DENICOLA, cited above in note 280, at 380-91; S. SHEMEL & M. KRASILOVSKY, THIS BUSINESS OF MUSIC 135-52 (rev. ed. 1971).

n295 See supra pp. 1713-14.

n296 See supra pp. 1726-27.

n297 Cf. R. POSNER, supra note 204, at 274 (discussing the cost savings such systems make possible); Note, Blanket Licensing: A Proposal for the Protection and Encouragement of Artistic Endeavor, 83 COLUM. L. REV. 1245, 1259-62, 1271-75 (1983). Depending, inter alia, on the proportion of the market it controls and its effects on the "output" of copyright licenses, such an organization might be held to violate the antitrust laws. Cf. Broadcast Music, Inc. v. Columbia Broadcasting Sys., 441 U.S. I (1979) (holding that a blanket licensing agreement does not constitute a per se violation of the Sherman Act, but may be subject to scrutiny under the rule of reason); Columbia Broadcasting Sys. v. American Soc'y of Composers, Authors & Publishers, 620 F.2d 930 (2d Cir. 1980) (holding that the particular blanket licensing scheme in question was reasonable and lawful), cert. denied, 450 U.S. 970 (1981);note, supra, at 1262-71 (reviewing possible bases of liability). The pretext for judicial intervention created by such a violation would create another opportunity for enhancement of economic efficiency. If a court were able to limit the size of the licensing fees charged by the organization, it could use the principle discussed above, see supra p. 1723, to improve the incentive/loss ratio. Cf. R. BROWN & R. DENICOLA, supra note 280, at 382-86 (reviewing the consent decrees under which ASCAP operates, some aspects of which provide for judicial review of the "reasonableness" of the royalties it charges).

n298 In making this point, I follow Professor Gordon, who argues persuasively that a court presented with a fair use defense should try to determine whether, if the defense were rejected, the parties would develop a "clearinghouse system to simplify the process of purchasing permission and thus allow a market to function." Gordon, supra note 185, at 1620.

n299 See Emerson v. Davies, 8 F. Cas. 615, 619 (C.C.D. Mass. 1845) (No. 4436) (Story, J.); R. POSNER, LAW AND LITERATURE: A MISUNDERSTOOD RELATION ch. 7 (forthcoming 1988).

n300 The development of photography in the late nineteenth and early twentieth centuries and the development of video games in the 1970's may be examples of such circumstances.

n301 It appears that the effectiveness of such an "infant industry" approach to the interpretation of the fair use doctrine would be maximized if courts were able to keep secret their plan to reduce the level of copyright protection after the initial surge of creativity had passed. However, leaving aside the problem of preventing premature "leaks" of the courts' intentions, the diminution of other artists' confidence in the stability (and decency) of the legal system that would result from implementation of such a strategy would more than offset the strategy's short-term benefits. For fuller exploration of the question of artists' confidence in "the system," see section IV.D.2 above.

n302 For a description and criticism of the state of the law pertaining to this problem, see Menell, cited above in note 198, at 1353-67.

n303 If the benefits to consumers of transformative uses of copyrighted materials are so great, why cannot the persons who engage in such uses charge consumers enough to enable them to purchase licenses from the copyright owners? If T. S. Eliot wants to incorporate in The Waste Land passages from other authors' works, see R. POSNER, supra note 299, at 38-40, why not let him buy permission from those authors to use their language? The result would be to increase the price of The Waste Land, but not to diminish the stock of intellectual creations. For three reasons (operative to different degrees in different contexts), voluntary transactions of this sort cannot be relied upon to facilitate creative uses. First, various circumstances are likely to prevent the transformative user from charging consumers the full value to them of his work -- for example, the infeasibility of engaging in perfect price discrimination and the impossibility of preventing some unauthorized copying of his work. Second, a transformative user (like Eliot) who wishes to incorporate passages from many prior works would incur enormous costs in negotiating licenses with all of the copyright owners in question. Third, many copyright owners would refuse to sell permission to make use of their works for any price. The third problem is especially serious in the context of parody. See Note, The Parody Defense, supra note 13, at 1397 n.12 (arguing that, "because of parody's caustic nature, a parodist often would not be able to secure an original author's permission through normal licensing procedures"). But cf. Neterville, Copyright and Tort Aspects of Parody, Mimicry and Humorous Commentary, 35 S. CAL. L. REV. 225, 233 (1962) (describing one situation in which a copyright owner was willing to license a parody of his work for a rumored fee of $100,000).

n304 See supra pp. 1710-12. For the purpose of this reexamination of the case, assume that, if the activity were declared unfair, the parties would negotiate a licensing arrangement of the sort described in section B, not a more sophisticated agreement of the sort described in the preceding subsection.

n305 See supra note 244.

n306 The foregoing modification of the analysis presented in section B might not be the end of the tale. If board games or their components were themselves copyrightable, see supra note 224, perhaps the diminution in the compensation paid to their inventors by a rule holding unauthorized uses of detective stories unfair could be offset in whole or in part by an adjustment in the rules governing unauthorized uses of board games. If so, the judge might be able to improve the newly revised ratio associated with Barker's behavior by making a change in the fair use doctrine as applied to games. The procedure the judge would have to employ to ascertain the most efficient combination of rules is sufficiently complex -- and sufficiently dependent upon information likely to be unobtainable -- as not to merit presentation here.

n307 In a paper that came to hand as this Article was going to press, Judge Posner and Professor Landes develop a formal economic model that emphasizes the costs associated with preventing authors from making creative use of copyrighted materials. See W. Landes & R. Posner, An Economic Analysis of Copyright Law (draft of April 13, 1988). The authors' recommendations regarding the shape of copyright law in general and the fair use doctrine in particular are consistent with the argument advanced here.

n308 Cf. supra section II.B.5 (discussing the rarity of concensus regarding the "propriety" of a given unauthorized use of copyrighted materials).

n309 For typical statements of that opinion, see Editorial, But Is It Piracy?, Wash. Post, Jan. 23, 1983, at C6, col. 1 (arguing that the Court should reject the studios' pleas); Shales, I'll Tape Tomorrow; And So Will You, Thanks to the Court, Wash. Post, Jan. 18, 1984, at B1, col. 1 (approving the outcome of the case).

n310 If the rules proved to be exactly as the actor suspected, the time and effort spent ascertaining them would constitute a deadweight loss.

n311 Assuming the project (purchasing a new car, for example) would have made the actor better off, the decision not to undertake it also results in an efficiency loss.

n312 In section B, for example, it was assumed that Americans' tastes for detective novels equivalent in quality to More Adventures would not be affected significantly by the frequency with which they purchased and read such novels.

n313 The relevant literature is reviewed in March, Bounded Rationality, Ambiguity, and the Engineering of Choice, in RATIONAL CHOICE 142-70 (J. Elster ed. 1986).

n314 The scholarship pertaining to phenomena of the sorts described in this paragraph is insightfully canvassed in Sunstein, Legal Interference with Private Preferences, 53 U. CHI. L. REV. 1129, 1145-66 (1986).

n315 On the possibly addictive character of jogging, see Taylor, Sallis & Needle, The Relation of Physical Activity and Exercise to Mental Health, 100 U.S. DEPT. OF HEALTH AND HUMAN SERVS., PUB. HEALTH REP. 195, 199 (March/April 1985).

n316 Economists refer to attitudinal shifts of this sort as "endogenous changes in tastes." See e.g., Yaari, Endogenous Changes in Tastes: A Philosophical Discussion, in DECISION THEORY AND SOCIAL ETHICS: ISSUES IN SOCIAL CHOICE 59 (H. Gottinger & W. Leinfellner eds. 1976). Tastes susceptible to modification in this way are sometimes contrasted with "exogeneous" preferences. See, e.g., March, supra note 313, at 153 (defining "exogenous tastes" as tastes that, "by whatever process they may be created, are not themselves affected by the choices they control"). As the example of chess suggests, it is difficult in many circumstances to discern whether an actor's tastes have been changed by engaging in an activity or whether he has merely learned that he has certain tastes and that they can be satisfied by continuing to engage in the activity. Because the two effects require similar modifications of the economic analysis of the fair use doctrine, the distinction is not pursued here.

n317 The more important reasons are: (i) my total wealth is greater if I am entitled to the wine or clean air and am considering selling it than if I am being asked to purchase it, and the greater my wealth, the less value I place on a dollar and therefore the more dollars I am willing to spend for a given entitlement, see, e.g., R. POSNER, supra note 204, at 15; (ii) possession of the wine or clean air will sensitize me to the pleasures it can provide; and (iii) my psychological makeup prompts me to prefer received income to opportunity income, see Kelman, supra note 196, at 678-81; but cf. Spitzer & Hoffman, supra note 196, at 1189 (arguing that Kelman fails to substantiate his thesis).

n318 For discussion of this general phenomenon, see Sunstein, cited above in note 314, at 1150-52 (referring to it as an "endowment effect"); Thaler, Illusions and Mirages in Public Policy, 73 PUB. INT. 60, 64-66 (1983); and Thaler, Toward a Positive Theory of Consumer Choice, 1 J. ECON. BEHAV. & ORG. 39, 43-47 (1980). A related aspect of the dependence of tastes on behavior is the phenomenon of so-called "posterior preferences." We often "find meaning and merit in our actions after they are taken and the consequences are observed and interpreted." March, supra note 313, at 160. In other words, through rationalization, we make our preferences conform to our conduct rather than vice versa.

n319 To illustrate: X decides not to take up chess because she suspects that she would come to love it and consequently would devote an inordinate amount of time to the game. Y enters law school rather than business school because, although he predicts that he would be happier ten years hence as a businessman than as a lawyer, he currently believes that law is a more socially responsible profession and he does not wish his sense of civic duty to atrophy. For insightful discussions of decisions like these, see J. ELSTER, ULYSSES AND THE SIRENS: STUDIES IN RATIONALITY AND IRRATIONALITY 76-85 (rev. ed. 1984), and Schelling, Enforcing Rules on Oneself, 1 J.L. ECON. & ORG. 357 (1985).

n320 For examples and thoughtful discussion of the practice, see J. ELSTER, cited above in note 319, at 84-85; Andenses, The Moral or Educative Influence of Criminal Law, in LAW, JUSTICE, AND THE INDIVIDUAL IN SOCIETY: PSYCHOLOGICAL AND LEGAL ISSUES 50 (J. Tapp & F. Levine eds. 1977); and Sunstein, cited above in note 314, at 1137-38, 1140-42, 1153-54.

n321 See generally Panel Discussion: Art as a Public Good, 9 COLUM. J. ART & L. 143 (1985) (debate among Professors Ronald Dworkin, Frank Michelman, Robert Nozick, and Thomas Scanlon, moderated by Professor Judith Thomson).

n322 Assume the judge has undertaken an inquiry of the sort sketched in section IV.D.1 above and concluded that "play" is the appropriate category.

n323 See supra note 303.

n324 The "price" would be the price of admission if a parody were produced as a play, or the fee demanded by the studio producing a parody if it were presented on commercial television. The price would rise not only because of the constriction of the supply, but also because the cost of each parody would now include the fee demanded by the owner of the copyright in the work parodied.

n325 See J. ELSTER, SOUR GRAPES: STUDIES IN THE SUBVERSION OF RATIONALITY (1983); Sunstein, supra note 314, at 1135, 1138-39, 1146-50.

n326 It is not obvious that the aficionados will respond this way. Although preventing a person from satisfying his desire for a good may make him value it less (apparently in order to avoid the "cognitive dissonance" associated with frustration), it may prompt him to value it more -- a phenomenon sometimes referred to as the "the grass is always greener" effect, see Sunstein, supra note 314, at 1148, as a "counteradaptive preference,", see J. ELSTER, SOUR GRAPES, supra note 325, at 111-12, or as the "forbidden fruit is sweet" effect, see id. at 111.

n327 See Spitzer & Hoffman, supra note 196, at 1211 (discussing the impact of a declaration that a certain activity is illegal upon people's desire to pursue that activity). Again, this effect is far from inevitable. To some people in some cultures, a good or activity is more attractive if the government opposes it. See id. (discussing the possibility that young people may demand more marijuana because it is illegal). In the hypothetical case, the judge concludes that delegitimation of parody will reduce rather than enhance demand for it.

n328 The reason for the reduction in monopoly losses is that the diminution in demand for parody would reduce the consumer surplus foregone by depriving consumers access to parody. See Figure 1, supra note 201.

n329 It might seem that the diminution in demand for parody would reduce the reward copyright owners could reap (in the form of fees granting parodists permission to ply their trade) in direct proportion to the diminution of monopoly losses, thereby leaving the ratio unchanged. The reason this will not occur is that most of the "reward" to copyright owners caused by classifying parody as an unfair use derives from avoiding the injury to the reputation of their works caused by parodies, not from the fees they can charge parodists. That reward will not be altered by the reduction in demand for parody.

n330 The judge of course could not take for granted that the ratio associated with board games will remain unchanged. In this case, he concludes that the percentage of the market for board games composed of games derived from plays is so small that neither of the decisions he is considering will have a significant impact on the supply, price, or apparent legitimacy of board games.

n331 Two kinds of impact on popular tastes affect most clearly the design of an optimal fair use doctrine: (i) preference changes that alter the incentive/loss ratios associated with putatively infringing activities; and (ii) preference changes that increase or decrease the aggregate rewards available to creators of a particular art form and therefore require an expansion or contraction of the total package of legal entitlements accorded those creators. The parody/board game case recounted above, see supra pp. 1735-36, illustrates the first sort of impact. Changes of that type are likely to occur whenever the copyright owner's response to a declaration that a use is unfair is to deny permission to engage in it (rather than license it for a fee) and whenever a substantial portion of the monopoly losses associated with forbidding an activity consist of the relatively stable costs of running an organization like ASCAP or BMI, see supra pp. 1728-29. Sony involved many impacts of the second sort.

n332 Cf. B. ACKERMAN, RECONSTRUCTING AMERICAN LAW 71 & n.31 (1984) (arguing that the "Coasean analysis" must be expanded to take into account "how the law shapes social perception and evaluation through a complex process of education and indoctrination").

n333 Cf. J. ELSTER, SOUR GRAPES, supra note 325, at 110 n.3 (the author does "not know of any discussions in the economic literature of adaptive preferences" in the sense in which he uses the phrase). Professor Yaari's article contains a brief discussion of the legitimacy of "regulatory intervention" to control exchange processes contaminated by the phenomenon of endogenous changes in tastes. See Yaari, supra note 316, at 88-93. Unfortunately, that discussion bears at most a tangential relationship to the question presented in the text.

n334 See supra p. 1699.

n335 The hypothetical case often used to illustrate this point is as follows. Assume that X, who now lives in the country, is well aware of the pleasures the city has to offer but still likes country living better than city living and does not wish to move. Assume further that, if forced to move to the city, X's tastes would change and she not only would come to prefer city living but also would be happier than she had been in the country. Should a policymaker devise some way to compel X to move to the city? Professor Sen at one time seemed to suggest that the answer is yes, see A. SEN, EMPLOYMENT, TECHNOLOGY, AND DEVELOPMENT 53-54 (1975), but has since disavowed that view, see Sen, Plural Utility, 81 PROC. ARISTOTELIAN SOC'Y 193, 213-14 n.41 (1980). Professor Roemer, in an unpublished paper, argues that forcing X to move would entail a degree of "paternalism" or "authoritarianism" that "will appeal to only a limited number of people." J. Roemer, Optimal Endogenous Preferences 9, 12 (1985) (unpublished draft on file with the Harvard Law School Library). He argues, however, that if the policymaker can devise a procedure for inducing X gradually and "voluntarily" to change her mind (e.g., by offering her a few amenities of city life (which she likes and selects even on the basis of her country tastes), the consumption of which alters her tastes, then offering her a few more amenities, etc.), he ought to do so. See id. at 10. Professor Elster disagrees. See J. ELSTER, ULYSSES AND THE SIRENS, supra note 319, at 81-83.

n336 For a general discussion of the distinction between first- and second-order desires, see Frankfurt, Freedom of the Will and the Concept of a Person, 68 J. PHIL. 5 (1971).

n337 Dworkin, Liberalism, in PUBLIC AND PRIVATE MORALITY 113, 125 (S. Hampshire, ed. 1978) (emphasis in original). For an argument that taking into account "preferences about preferences" also has the advantage of allowing us to screen out people's nastiest desires and therefore makes welfare economics a less easy target for its critics, see Goodin, Laundering Preferences, in FOUNDATIONS OF SOCIAL CHOICE THEORY 75 (J. Elster & A. Hylland eds. 1986).

n338 See Dworkin, supra note 337, at 127; see also R. DWORKIN, supra note 169, at 441 n.19.

n339 To the extent that persons' second-order preferences incorporate not just desires regarding the kinds of tastes they themselves wish to have, but also convictions regarding the sorts of tastes everyone ought (or ought not) to have, the proposed procedure would also run afoul the liberal principle that "external preferences" should not be taken into account when making collective decisions. See, e.g., R. DWORKIN, supra note 139, at 235-36.

n340 See Sunstein, supra note 314, at 1145-46.

n341 See Dworkin, supra note 337, at 143 ("[S]ince the preferences people have are formed by the system of distribution already in place, these preferences will tend to support that system, which is both circular and unfair."); Roemer, Rational Choice Marxism: Some Issues of Method and Substance, in ANALYTICAL MARXISM 193 (J. Roemer ed. 1986).

n342 Professor Sunstein, whose writings have stimulated much of my thought in this area, responds somewhat differently to the perception that, in many contexts, the arguments in favor of deferring to popular preferences are weak. His proposal is that we should ignore or override those preferences and use the law to inculcate different tastes when and only when the "autonomy" of the affected persons would be enhanced. See Sunstein, supra note 314, at 1136. Cf. J. ELSTER, SOUR GRAPES, supra note 325, at 135-40 (also attempting to develop and rely upon the notion of "autonomous" desires).

Sunstein's reluctance to define "autonomy" makes it difficult to assess his proposal. My tentative judgment is that the enterprise, to the extent it attempts (by referring to an ostensibly neutral notion of freedom) to avoid the highly controversial task of articulating a theory of better and worse tastes (i.e., a vision of the good life), cannot succeed. Final judgment on that question must await Sunstein's elaboration of his argument. See Sunstein, Disrupting Voluntary Transactions, 29 NOMOS (forthcoming 1988).

n343 For an explanation of the need for such information, see p. 1715 above.

n344 Cf. Kaplow, supra note 210, at 1842-43 (expressing pessimism regarding the capacity of courts to develop an efficient resolution of the tension between patent and antitrust law); Perlman & Rhinelander, Williams & Wilkins Co. v. United States: Photocopying, Copyright, and the Judicial Process, 1975 SUP. CT. REV. 355, 373-79 (arguing that the "hard evidence"necessary to evaluate a particular invocation of the fair use doctrine on economic grounds is not available); Priest, What Economists Can Tell Lawyers About Intellectual Property, in THE ECONOMICS OF PATENTS AND COPYRIGHTS, supra note 185, at 19-24 (arguing that the answer to the question is "very little," primarily because the empirical data necessary to make intelligent recommendations is altogether lacking).

n345 See supra p. 1707.

n346 See supra pp. 1700, 1715.

n347 Or, conversely, any predictable reduction in creators' welfare caused by permitting the use should be counted as a "harm."

n348 724 F.2d 1044 (2d Cir. 1983), cert. denied, 469 U.S. 823 (1984).

n349 Id. at 1050 (emphasis added).

n350 Id. at 1051 (emphasis in original).

n351 The denominator of the fraction would be the efficiency losses that resulted when consumers, denied immediate access to the language used by Consumer Reports in its review, made purchases they would not have made had they been given easier access to that language.

n352 Another recent illustration of this common error is the decision of the the Ninth Circuit in Hustler Magazine v. Moral Majority, Inc., 796 F.2d 1148 (9th Cir. 1986). The Moral Majority, as part of a fundraising campaign, sent to some of its members letters which included photocopies of a copyrighted obscene parody (featuring Rev. Jerry Falwell) which had been published by Hustler Magazine. Hustler brought a copyright infringement action against the Moral Majority, which in turn invoked the fair use doctrine. A bemused panel of the court of appeals found in favor of Moral Majority, largely on the ground that "Hustler's creative incentives are not decreased [by the mailing] because the Defendants are profiting from an activity that Hustler could not have taken advantage of." Id. at 1156. One judge dissented, criticizing his colleagues for their apparent willingness to take judicial notice of the fact that "Moral Majority . . . members would probably not be counted among Hustler's readers." Id. at 1158 n.1. (Poole, J., dissenting). The dissent did not, however, take issue with the manner in which the majority calculated the potential "harm" to Hustler. All three judges thus deemed irrelevant the fact that the purpose of Moral Majority's campaign (from which it gained over $700,000) was to help finance a suit by Falwell against Hustler for libel, invasion of privacy, and infliction of emotional distress. If, as seems certain, Moral Majority's unauthorized use of the parody increased its litigation fund and thereby enabled it to mount a more effective suit, cf. Hustler Magazine v. Falwell, 108 S. Ct. 876 (1988) (describing the outcome of the litigation), the net impact may well be to reduce the eagerness of Hustler and other magazines to publish parodies of public figures in the future. Whether that result would be unfortunate is debatable, but the argument that this sort of unauthorized use of copyrighted material would have no effect on "creative incentives" is plainly incorrect.

n353 Typical of articles of this sort is Professor Light's discussion of parody, the thesis of which is: "The central consideration in deciding whether a parody infringes a copyright will be whether it impinges upon the copyright holder's legitimate interests by competing with his work in a market in which that work would ordinarily be salable." Light, supra note 185, at 634. Articles taking similar tacks include Comment, Photocopying and Fair Use: An Examination of the Economic Factor in Fair Use, 26 EMORY L.J. 849, 865-71 (1977), which argues that "[t]he principal question in fair use cases is whether the demand for plaintiff's copyrighted work is partially satisfied by defendant's use" and therefore that photocopying should be permitted whenever it does not impair the demand for the "originals"; Keon, Audio Home Recording: Canadian Copyright Implications, in THE ECONOMICS OF PATENTS AND COPYRIGHTS, cited above in note 185, at 157, 171, which takes for granted that the law should be concerned with home recording only if the practice reduces rather than increases record sales; and Liebowitz, cited above in note 185, at 191, which makes a similar assumption about the practice of photocopying professional journals. Professor Nimmer criticizes such narrow definitions of injury, but nevertheless contends that the courts' inquiries should be bounded by a "functional test." See 3 M. NIMMER, NIMMER ON COPYRIGHT § 13.05[B], at 13-84 (1987) ("If the defendant's work adversely affects the value of any of the rights in the copyrighted work, . . . the use is not fair even if the rights thus affected have not as yet been exercised by the plaintiff. But if regardless of medium, the defendant's work . . . performs a different function than that of the plaintiff's, the defense of fair use may be invoked." (footnotes omitted)) The rationale for thus constricting the category of legally cognizable harm is not apparent.

n354 See supra pp. 1709-10.

n355 See supra pp. 1710-12, 1726-28. Excluded from those discussions was the possibility that Barker's market power derived from the originality of the board games it developed. The significance of "creativity" of that sort is considered in section IV.D.5 and the subsection immediately following this one.

n356 See supra p. 1728.

n357 In circumstances in which the infringer enjoys significant market power because of the originality of his derivative work and in which both copyright owner and infringer are sophisticated bargainers, efficiency losses of the sort discussed in this section would form the bulk of the denominator of the ratio. In other words, the lesson stated in the last paragraph of the immediately preceding subsection and the lesson stated in this subsection will sometimes pull in opposite directions.

n358 Professor Gordon acknowledges the relevance of the "productivity" of the putatively infringing activity only insofar as it may affect either the likelihood of adverse impact on the income of the copyright owner or the likelihood that transaction costs will prevent the copyright owner and infringer from reaching an efficiency enhancing agreement. See Gordon, supra note 185, at 1653-54. Contending that those effects will be modest, Gordon takes the position that "the productive status of a user is at best a secondary indicator" of the economic considerations upon which the availability of the fair use defense ought to depend. See id. at 1601. The analysis offered here suggests that creativity should play a more substantial role in the doctrine.

n359 See supra section II.D.

n360 See, e.g., supra pp. 1725-26 (discussing the advantages of compulsory licensing systems).