Last major update 10-26-01



L.H.O.O.Q.--Internet-Related Derivative Works

Use NS 6.1+ or I.E. 5+, 800 x 600 screen recom.
This page will not work without JavaScript and HTML-4 <IFRAME> capability


[Cont'd from previous sitepage]

. . . A second work might be a derivative work based on an earlier work, but it might not be infringing if the elements that it took from the earlier work were not copyright-protected elements (for example, suppose that the elements taken are facts, ideas, or mise-en-scene). Even if the taking is infringement, the taking might be licensed, fair use, or otherwise excused or privileged. Arguably, the three things -- derivative work, infringing derivative work, copyright-liable derivative work -- are successive subsets. Our focus belongs on the second and third subsets, and on the seond subset only because the analysis must get past that subset to reach the third one.

What elements distinguish an infringing derivative work from one that is merely based on an earlier work without infringing its copyright? One conceivable test is whether the second author adds a quantitatively substantial amount of original material to the original work. Another is whether the second author takes a quantitatively substantial amount of material from the original work.

Neither of these possible tests is very useful. Virtually all observers would agree that L.H.O.O.Q. is a derivative work based on the Mona Lisa and that (assuming presence of copyright) it would infringe unless excused by a pertinent defense. The quantitatively small amount of material apparently present in Duchamp's added layer was disproportionately influential in transforming Mona Lisa into La Dame, L.H.O.O.Q. -- just as a small amount of catalyst may be effective to initiate a reaction. By the same token, a nearly total taking, as in Big Hairy Woman, will be excused if more pertinent factors so dictate. The proper test is not likely to turn on the quantum of material added or taken, although that factor may operate as a makeweight.

In recent years, since the Big Hairy Woman case, a great deal of attention has been given to transformative uses when evaluating whether the affirmative defense of fair use excuses an infringement. As indicated earlier in these pages, L.H.O.O.Q. manifestly transforms an icon into an object of ridicule. That transformation is a major factor in making L.H.O.O.Q. a derivative work rather than such an insubstantial variation on the original Mona Lisa as to be the original and therefore not open to possible guilt as an infringing copy. Why this is so is a reflection of the milieu or social setting that gave meaning to Duchamp's quantitatively small but qualitatively significant modifications of the original work. That is, the modifications made were such that they successfully mocked the cult of Jacondisme, facilitating Duchamp's intent to epater le bourgeois.


I am transformative.
Approach my event
horizon at your peril.
Unfortunately, the risk with that approach to analyzing transformation is that every case is sui generis. What does it mean for an addition to a work to be transformative in a qualitative sense other than for the observer to exclaim, "I know it when I see it"? Not only is that test unhelpful for purposes of prediction and rationalization of business activity, but it provides no clue for migrating the test from epater le bougeois to framing or otherwise changing the presentation of another person's Web page. As yet, no decision explains that. Moeover, much of the current commentary on transformation leaves that term without objective substantive content. Indeed, transformativeness is becoming one of the black holes of copyright law. It swallows anything that approaches it and no light emerges from it.

It may be possible to develop a liability test for Web page derivative works, however, by borrowing or extrapolating from the very limited case-law developed so far in nearby copyright fields. Such a test might borrow from the more general analysis of fair use to distinguish liability-attracting derivative works from privileged ones. The test will look to whether the supposedly transformative or privileged use by the defendant provides the public with a benefit not previously available to it or instead simply provides the public with an alternative channel for utilizing the original content, in effect in competition with the oiginal content provider. Perhaps more generally one might speak of instead simply providing an alternative means of economically exploiting the original content.

The decision in the Arriba Soft case is particularly suggestive of the direction that such a test might take. That case involved a wholesale preparation of reduced-size ("thumbnail") GIF and JPEG images from virtually all Web sites on the Net. The purpose was to provide a search engine, so that users could search the Internet for images of a given kind. Users of the search engine would determine whether to go to a site to examine or download an image by examining thumbnail versions of the various images that the search engine had gathered. In the court's view, the defendant provided the public with a very useful new service that previously had not been available, and the thumbnails were a necessary element for providing the service. At the same time, the thumbnail images did not compete with the ordinary commercial use of the underlying image works. For these reasons, the preparation of these derivative works and the defendant's related conduct did not violate any interest that copyright law protected. Rather, the conduct was fair use.

The Arriba Soft decision is more suggestive of public policy considerations than it is a restatement of established copyright precedents. For one thing, the case's fact pattern limits it to aspects of copyright law not generally applicable to derivative-work controversies. Moreover, the defendant conceded reproduction of copies of the original works. (The court states that the defendant "does not dispute it reproduced and displayed Plaintiff's images in thumbnail form without authorization.") Apparently, the defendant copied the image files, processed them to make thumbnails, stored the thumbnails on disk for display to users via the search engine, deleted the original image files (as occurs for intermediate copies in reverse engineering), and stored and provided users of its search engine with hyperlinks to the original image files at their original sites. Users then displayed the thumbnails on their computer monitors in the course of using the image search engine. Users may also have linked to the hyperlinked sites or have downloaded the images from such sites.

Apart from reproducing the original image files to make the thumbnails and storing and utilizing the thumbnails as described, the defendant did not itself reproduce copies of the original images or distribute them. Offering links to them is not reproduction or distribution. Whether the stored thumbnail images are reproductions of copies or just preparations of derivative works is debatable. It may depend on how close the thumbnails are to the originals in their expressive aspects. (Is a copy of Rauschenberg's Erased De Kooning a reproduction of the original work or is it a copy of a derivative work based on the original? Is it neither one?) In affirming the district court's conclusion that the use was fair, the court of appeals suggested that the much lower resolution of the thumbnails prevented their use for artistic purposes in competition against the original pictures. This both made the purpose and character of the defendant's use different from the copyright owner's and prevented supersession of the copyright owner in the marketplace. The discrepancy between the works tends to make the defendant's conduct more the preparation of a derivative work than the reproduction of a copy, although the court of appeals' analysis treated the case in substantial part as one of copying (public display of a copy of the work).

If it were technically feasible to provide search engine users with evanescent thumbnails "on the fly," as in a search engine's tabulation of the results of a given search, the thumbnail works could even more definitely involve only preparations of derivative works and not reproductions of fixed copies (stored other than in users' temporary memories). But the realtime requirements of processing image data in response to search requests requires use of stored image data, which in turn requires reproduction of expressive aspects of the originals in substantially permanent memory. Change that aspect of the thumbnail fact pattern and this case could be a pure vanilla derivative work case. That images, rather than text, are involved may subtlely affect the nature of the infringement. The conduct in Arriba Soft implicates the reproduction and public display clauses of 17 U.S.C. § 106, and potentially the distribution clause as well. The case may therefore be distinguishable on several grounds from other Internet uses of other persons' Web page content. Nonetheless, the treatment of transformation in Arriba Soft may well suggest a legal theory for framing and linking cases with a less special fact pattern.

The significant aspect of Arriba Soft, for the present purpose, is its balance of fair use factors to find the defendant's conduct excused from copyright liability. The court compared the defendant's purpose (using the images for searching the Internet) and the plaintiff's purpose (using images per se, for illustration or decoration). Because the later use was quite different and was very useful to the public, the later use gave the public a substantial new benefit. The court said that such uses are transformative. In effect, the district court's opinion equates transformative with novel and substantially beneficial to the public.

The Arriba Soft court's approach to transformative fair use has not gone without criticism. Professor Jane Ginsburg (Internet link) rejects this analysis and maintains that "[t]here is nothing transformative about taking an image and reproducing it in full." Moreover, she denies that the thumbnail is "incorporated into a new work that builds or comments on the nature or qualities of the copied work."

The first point is more an assertion of a conclusion rather than a demonstration of its correctness. Whether the changes made by a reduction in size and elimination of detail (perhaps of color gradations as well), and the placement of the reduced-size image in an array of other images concerning the same subject matter, effect a transformation is a question that can be decided only after making a factual inquiry into the consequences of the action. If the district court erred in this assessment, that should be shown rather than simply asserted.

More fundamentally, one might question what difference it makes if the black-hole epithet transformative is mangled somewhat in the court's opinion. One could appropriately reach the same result by saying, instead, that part of the purpose and character of the accused use, per § 107(1), is that it improves users' access to the Internet by providing them a new search function not previously available and this significantly benefits the public. The only reason for using the epithet transformative is rhetorical: to be able to rely on the Supreme Court's assertion in the Big Hairy Woman case that "the more transformative the new work, the less will be the significance of other factors, like commercialism, that may weigh against a finding of fair use." One could say, with at least equal rationality, that the more beneficial to the public the new use is, the less significant other factors will be. On the other hand, one must expect lawyers to flourish rhetorical tropes and to prefer reciting quotations to crafting stand-alone reasoning. So long as the result is not to make the worse appear the better cause, little harm is done and the process is tolerantly called making noises like a lawyer. The Arriba Soft court's use of the word "transformative" does little or no more harm than has already been done so far by use of that term, and there is no point carping at it. Indeed, one might well conclude that, since you cannot beat them, you would do best to join them.

The second point of Professor Ginsberg's analysis does not fully take into account that the defendant created a number of compilations of images each having a given subject matter. Compilations are works; section 101 says so. The compilation of a number of Internet images on a given topic builds on the underlying works to provide users with the opportunity to screen the images visually to find one or more of them that are suitable for a purpose of the user (say, the user wants a picture of a black hole to illustrate a metaphor that some copyright doctrine is a black hole, or wants a picture of Chaucer to show Chaucer as a notional copyright proprietor taking issue with someone else, such as a notional Professor Annotator.). Such a compiled work amply and usefully "builds on" the original work's "nature," assuming that copyright law requires that characteristic in a derivative work.

In any case, as will appear, it is not most important here that the Arriba Soft court properly resolved the merits of the issues before it. What matters most is whether its concept instructs us as to Internet-related cases about derivative works more like the alertbox demonstration and popup window advertisements than the thumbnails and image search engine of Arriba Soft. The purpose of the present inquiry is to determine whether we can usefully build on a legal standard for justifying or condemning a derivative use that asks if the derivative use provides the public with a new function, previously unavailable, and if the new function is highly beneficial to the public. Arriba Soft asks this by proposing an equation: highly beneficial to public = transformative. In this context, beneficial to public can probably be said to include, at a minimum, all of the functions that 17 U.S.C. § 107 lists in its preamble: criticism, comment, news reporting, teaching, scholarship, research, for they all are deemed socially beneficial. Those uses are in turn subsets of the generic category of promoting the progress of science (human knowledge), the constitutionally dictated purpose of copyright law. Conceivably, other uses can be brought within the scope of beneficial to the public, but without a statutory or constitutional standard it is uncertain what to make the measure of public benefit, and any ad hoc case by case analysis not tied to a statutory anchor can become problematic. Initially, therefore, it would be better to define public benefit in terms of what copyright law says is public benefit.

Accordingly, the proposed test to be extrapolated from Arriba Soft may be stated in the following terms:

A defendant that prepared a second, derivative work on the basis of a first work is excused from copyright liability if the second work
  • does not actually or potentially divert trade in the marketplace to itself from the first work;

  • provides the public with a previously unavailable, very useful function of the kind that copyright law exists to promote.

What is the appropriate connective? Must it be and? Will or suffice? Possibly, but that is a more difficult argument. It will need exploration. As will appear, when public benefit is present and diversion is absent, it will be an easy case in favor of the derivative use. When public benefit is absent and diversion exists, it will be an easy case for condemnation of the derivative use. When both public benefit and diversion exist, the case becomes difficult.

This emerging theory of derivative work liability for Internet related works may be further illustrated and analyzed in terms of a hypothetical case based on another recognized kind of derivative work -- the hyperlinked textual annotation. Consider this annotation of the immortal Professor Chaucer's work, the General Prologue to the Canterbury Tales. Let us assume that Professor Chaucer has thus far published or licensed it only in two forms. The original work is the left-hand column (Whan that Aprille..."), published circa 1400. He later licensed a derivative work shown at left, in the form of a two-column text, one column in Middle English and the other in Modern English. While this mode of presentation is somewhat awkward, until now Professor Chaucer has considered that it is not consistent with his business plan or is just not cost-effective to come forward with a more streamlined version. (He is immortal but not fully au courant.)

Professor Annotator is dissatisfied with the visual appearance and lack of convenience of this product. He considers that a more user-friendly product would be more consistent with the attention span and learning attitudes of his students. He now publishes the annotation shown below, which features mouseover tooltips as the means for communicating information to overcome the differences between Modern English and Middle English. A tooltip is a small popup window layered over a text page near a phrase to which it relates, much in the manner of the moustache gif of the L.H.O.O.Q. animation shown earlier.

This action leads Professor Chaucer to sue Professor Annotator for preparation of an unauthorized derivative work. (Assume, arguendo, that framing or linking to Professor Chaucer's Web site, where Professor Chaucer placed the Prologue, is used in some manner to avoid a claim of reproduction -- to leave a pure § 106(2) case.) One thrust of Chaucer's argument is that the annotation substitutes competitively for his own commercial work (the circa 1400 left-hand page shown above). In addition, Annotator's work competes even more substantially against Chaucer's licensed commercial derivative work (the right and left columns above that provide a line-by-line display of Middle English and Modern English in parallel columns). The argument of Annotator would be that his annotation with pop-up tool-tips, as shown here, provides the public with a beneficial facility not otherwise available. Annotator argues that his product is much more convenient and intuitive, and thus more useful to the public (here, students) than the licensed work, for purposes of instruction, scholarship, and criticism.

Chaucer responds that the use of pop-up tool-tips is an idea rather than expression, so that if the public wants this product he can adopt it without becoming liable to Annotator and he as copyright owner should have the exclusive right to supply the public's demand for pop-up tool-tip annotations relating to his underlying work. Annotator might respond that without competitive pressure Chaucer has insufficient incentive to innovate this way, to put the new product on the market, and to satisfy the public's needs. Accordingly, the public would not get the benefit of such advances if they are not deemed fair use. Chaucer would respond that copyright owners should have the absolute right to decide which derivative works will be placed on the market and which will not, as an inherent part of their exclusive right under § 106 and Art. III, § 8, cl. 8 of the Consititution.

That last is undoubtedly the strongest argument on Chaucer's side of the case and the most difficult issue to resolve. Support for this argument can be found in the Seinfeld trivia book case. In that case, the Second Circuit determined that a trivia book based on Seinfeld TV show episodes was a derivative work that the fair use doctrine did not shield from infringement liability. The court recognized that the copyright owner had "little if any interest in exploiting this market for derivative works." The court nonetheless concluded that "the copyright law must respect that creative and economic choice." Why must? Apparently on the ground that having a statutory monopoly necessarily implies that power. The court went on, quoting the district court, that it would not advance the arts to deny artists "their monopoly over derivative versions of their creative works merely because they made the artistic decision not to saturate those markets with variations of their original." That argument is highly questionable. The right to get courts to enforce such artistic decisions cannot be absolute. It is recognized, for example, that artists and their publishers do not license others to create derivative works that are parodies or biting criticisms. Surely, the owner of copyright in Pretty Woman would have made the artistic decision not to saturate the market with trash like Big Hairy Woman. So what? More generally, why assume that the public's interest will be maximized by according copyright owners such a veto power? Instead, we should ask whether giving or withholding that veto power will in the long run more encourage the progress of knowledge and the arts.

When the copyright owner does not want to create or license the derivative use, but the use or category of use (e.g., criticism) is beneficial to the public, it is recognized that the public will benefit more from allowing the use despite the copyright owner's wishes. (Whether some kind of royalty should be required is a further issue.) Saturating the market with variations is probably just a red herring. A derivative work, particularly a highly transformative one, is not in the same market as the original and is not a substitute for it. This branch of the case can therefore be ignored, at least for derivative works that greatly transform the original work.

The case is more difficult when the copyright owner says, as Professor Chaucer does in the hypothetical case, that he (unlike the copyright proprietor in the Seinfeld trivia book case) would presently get around to exploiting the market for this kind of derivative work, and that his exclusive right as copyright owner should give him the exclusive franchise to do so. Nonetheless, the response (above) of Professor Annotator that Chaucer's delay creates a credibility problem is appropriate. At the very least, why make the public wait in accordance with Chaucer's agenda? What if he takes a long time? Without competitive pressure, Chaucer has insufficient incentive to create the innovation, put it on the market, and give the public the beneficial enjoyment of the advance. Therefore, we need not assume (rather than worry over the facts) that the public will get the benefit of the advance within a reasonable time. Without gadflies like Annotator, the impatient public is unlikely to enjoy promptly the benefits of improvements on the original works, whether created by the gadflies or the original authors that they may buzz around.


- End of second page of LHOOQ -

[Proceed to next page of LHOOQ for additional material]