Fair or Foul? The Unanswered Fair Use Implications of the Google Library Project
The proposed Google Book Settlement is arguably one of the hottest copyright topics of 2009. After years of legal wrangling between Google and the Author’s Guild and publisher groups over Google’s unauthorized digitization, indexing, and display of portions of copyrighted books, the parties reached a tentative settlement in the fall of 2008. By proposing this settlement, which awaits court approval, Google skirts the 800-pound gorilla: whether its actions were protected under the fair use provisions of the Copyright Act of 1976. This question begs to be answered as the world becomes increasingly digital and users of technology continue to push the boundaries of the law to meet their own needs. Also, importantly, courts have never seen a fair use case of the same magnitude as the Google Library Project.
In an effort to provide that answer, this article presents an analysis of how the U.S. District Court for the Southern District of New York might have decided the fair use question through an analogy of cases the facts of which are arguably comparable to Google’s activities here. It is important to note that this question relates to Google’s pre-settlement activities of scanning, online indexing, and displaying snippets of text—not the more ambitious Book Rights Registry and other proposed terms that developed during the course of settlement discussions.
Google’s Grand Plans, Legal Roadblocks, and Settlement
In December 2004 Google announced the Google Library Project in which Google intended to digitally scan entire printed books from the collections of five libraries (University of Michigan, Harvard University, Stanford University, Oxford University, and New York Public Library), store them on Google’s servers, and index and display snippets of those books online. Although Google would not charge a fee to search its online archive, Google intended to derive revenue through banner advertising.
In September and October of 2005, the Author’s Guild and a coalition of publishers separately sued Google in the U.S. District Court for the Southern District of New York for copyright infringement based on the Google Library Project. The suits were eventually consolidated.
Copyright law is governed by the Copyright Act of 1976, 17 U.S.C. § 101 et al. Section 106 of this law enumerates six exclusive rights enjoyed by the owner of copyright. By digitally scanning copyrighted books without permission for the purpose of creating an online searchable database of those works, Google’s Library Project infringes the rights of reproduction,4 distribution,5 and public display6 held by copyright owners. From the beginning, Google has defended its actions under the fair use provision found in 17 U.S.C. § 107. This provision protects unauthorized uses of a copyrighted work for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research.
The copyright statute enumerates four factors that are used to determine whether the use of a work is a fair use:
1. the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes;
2. the nature of the copyrighted work;
3. the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and
4. the effect of the use upon the potential market for or value of the copyrighted work.
These factors are neither rigid not exhaustive, allowing courts to use their discretion to consider other relevant information. This practice was “warmly encouraged” by Congress when it amended § 107 in 1992.9 Some examples of additional factors to consider include the defendant’s good or bad faith10 and whether the defendant distorted the meaning of the original work.
There have been numerous cases where these factors have been examined in the face of a defendant’s fair use claim, and out of the many cases, those that are most analogous to the Google Library Project fall into three fact patterns: (1) Internet search engines that aggregate online content; (2) online music sites that share and distribute music files that have been converted from CDs to digital files; and (3) retransmission of content in a new medium. These fact patterns are relevant to the Google Library Project because they all involved infringement on the Internet or in a medium different than the one intended by the copyright owner, and in each fact pattern the defendant engaged in systematic infringement rather than isolated instances. Cases falling into these fact patterns are analyzed below. They allow one to draw conclusions as to which side of the fair use coin Google would end up on had the case been fully litigated.
First Factor: Purpose and Character of the Use
When analyzing the first factor, the two most important considerations are (1) whether the use is for a commercial or noncommercial purpose and (2) whether the use is “transformative.” The commercial/noncommercial question has created misapplications of the fair use doctrine because of false assumptions that have arisen over the years (e.g., educational noncommercial uses are always allowed;12 entertainment commercial uses will not be allowed). The reality is that a commercial use is less likely to be considered a fair use, but what is more important is whether the use is transformative. A transformative use is one that creates a new purpose or meaning of the work that was not contemplated by the owner of copyright. Courts have held that the more transformative a work, the less important the other fair use factors become.
Two major decisions arising out of the Ninth Circuit upheld the right of Internet search engines to display copyrighted works without the owner’s permission largely in part because of the transformative use arguments set forth by the defendants. In the first case, Kelly v. Arriba Soft, defendant Arriba Soft pulled 35 of plaintiff Kelly’s copyrighted photographs from online sources and displayed thumbnail images of these photographs in the defendant’s search engine without permission. When the images appeared in the search results, an in-line link displayed the full image and gave the user the impression that the image was on Arriba Soft’s servers, when it actually resided on Kelly’s server.17 Kelly sued and Arriba Soft defended on the grounds of fair use.
In holding that Arriba Soft’s use was fair, the determining factor in the Ninth Circuit’s decision was the transformative nature of Arriba Soft’s use of the images in its Internet search engine.18 The court ruled that even making an exact unauthorized copy of a work may be transformative so long as the copy serves a different function than the original work.19 Citing Campbell v. Acuff-Rose Music, Inc., 20 it said that the more transformative the new work, the less important the other factors, including commercialism, become.21 Because the purpose of Arriba Soft’s use was to improve access to images on the Internet, the court deemed it sufficiently transformative and a fair use.
Four years later, the Ninth Circuit upheld Arriba Soft in a similar case, Perfect 10, Inc. v. Amazon.com. In this case, the plaintiff, Perfect 10, Inc., was an operator of a subscriber-based website that sold access to images of nude models. Perfect 10 sued Google and Amazon.com separately for the unauthorized display and distribution of copyrighted Perfect 10 images that appeared as thumbnails in Google’s image search engine. The thumbnails linked to third-party sites where the images were After years of negotiations, the parties announced on October 28, 2008, a tentative settlement that would allow Google to continue its digitization efforts. Under the settlement, the following obligations and conditions were agreed upon: The settlement class automatically includes all copyright owners who hold U.S. copyright registrations for printed books and “inserts” prior to January 5, 2009; parties wishing to opt out must affirmatively do so. Google will be allowed to continue its efforts to scan, digitize, and include books in an electronic book database; sell advertising on pages containing digitized books; and, with the cooperation of copyright owners, sell institutional subscriptions to the electronic book database and sell individual online access to books. Copyright owners who remain in the class will receive 63% of all revenue generated from Google’s commercialization efforts related to the Library Project. Google also will make payments to copyright owners whose books were digitally scanned without permission if they remain in the class and submit a timely claim. At the time the settlement was announced, it was estimated that each legitimate claim will entitle the copyright owner to approximately $60 per book. Google also will be creating and funding a Books Rights Registry that it envisions as an opportunity for participating copyright owners who remain in the settlement class to commercialize their books via the Internet. In exchange for participating in the settlement, copyright owners release all claims of copyright infringement against Google and the libraries cooperating in the digital scanning efforts. The settlement was preliminarily approved on November 17, 2008, but the terms are by no means final. Parties have until September 4, 2009, to decide whether to opt out of the settlement, and the court will hold a fairness hearing on October 7, 2009, for remaining class participants to voice their objections or concerns with the settlement terms. These dates were extended by approximately four months after a rising tide of opposition among various interest groups and a U.S. Department of Justice inquiry regarding possible antitrust concerns related to the proposed Books Rights Registry. posted without Perfect 10’s permission. The cases were eventually consolidated. As with Arriba Soft, the Ninth Circuit held that use of thumbnail images in a search engine served a transformative public benefit by promoting access to information on the Internet as an electronic reference tool, and this public benefit outweighed Google’s superseding and commercial uses of the thumbnails.
The transformative use argument has been less successful in cases involving unauthorized use of music online. In both A & M Records, Inc. v. Napster, Inc. and UMG Recordings v. MP3.com, Inc., the courts were not persuaded that the unauthorized use by defendants was transformative. In A & M Records, Inc. v. Napster, Inc., the Ninth Circuit struck down Napster’s fair use defense to the peer-to-peer MP3 file sharing that Napster users engaged in during the late 1990s. Users would burn copies of their CDs as MP3 files and then make them available to other users to search and download using Napster’s software application.Turning to the first fair use factor, the court held that Napster’s use of plaintiff’s copyrighted music was not transformative because the content was merely retransmitted in a different medium. It also held that Napster’s use was commercial given that repeated unauthorized copying was done to save the expense of purchasing authorized copies.
In UMG Recordings v. MP3.com, Inc., defendant MP3. com copied tens of thousands of store-bought music CDs onto its servers for its subscribers to access and listen to from any computer with an Internet connection. MP3.com labeled this activity “space shifting” (as opposed to “timeshifting” as referenced in the U.S. Supreme Court case Sony Corp. of America v. Universal City Studios, Inc.). Before MP3.com’s subscribers could access the music online, they had to demonstrate that they already owned a copy of a particular CD (by inserting it into their computer and letting defendant’s site verify that it is a purchased copy) or that the subscriber had purchased a copy through one of defendant’s preferred online retailers. The Southern District of New York did not buy the space-shifting argument as a transformative use worthy of fair use protection and stated that merely retransmitting the original work in a different medium may be innovative but is not in itself transformative.
Even when a defendant articulates a different purpose for its use of the infringed work, courts sometimes have difficulty excusing the infringement as a fair use. In the Second Circuit case Infinity Broadcasting Corp. v. Kirkwood, defendant Kirkwood operated a paid service—“Dial-Up”— whereby subscribers could listen to plaintiff Infinity Broadcasting’s radio broadcasts from other cities via a telephone call-in number. Dial-Up was marketed to radio stations, advertisers, talent scouts, and others in the entertainment and advertising industry for purposes such as “auditioning on-air talent, verifying the broadcast of commercials, and listening to a station’s programming format and feel.” When Infinity sued for copyright infringement, Kirkwood defended on the grounds that his use was fair because it was for a purpose (i.e., advertising and talent scouting) that differed from that of Infinity (i.e., general entertainment). Even though the court acknowledged that the different purpose is an important consideration of the first factor of the fair use analysis, it said that the more significant inquiry is whether the use was transformative. In the end, the court found that Kirkwood’s retransmission of Infinity’s radio broadcasts over telephone lines was given no “new expression, meaning, or message” because the character of the original broadcasts remained unchanged; as a result, its use was not transformative. In these cases, the differences in the courts’ rulings appear to be whether the works were merely retransmitted in a different medium for the same purpose as the copyright owner intended, or whether the defendants articulated a satisfactory new purpose and changed the work in some way to fit that purpose. Based on these rulings, the following conclusions can be drawn: reducing an image to a thumbnail is an allowed use if that thumbnail is used to promote the spread of information on the Internet, but simply using a sound recording in a different medium without changing the recording in any way is not transformative, and, even when a defendant demonstrates a new purpose for the copyrighted work, if that defendant has not changed the work in any way, courts will hesitate to deem that use transformative.
As applied to the Google Library Project, Google would argue that providing an online searchable database of millions of books (including many that are out of print) available to the world is highly transformative because such a use promotes research and knowledge. The public benefit derived from widespread access to a searchable book database is arguably greater than the benefit that results from search engines that display thumbnail images of photographs (e.g., the Arriba Soft and Perfect 10 fact patterns). It should be noted that when a use is transformative, it can outweigh other factors such as whether the infringer is a commercial entity and whether the entire work has been used.
In contrast, the rights holders would argue that Google’s efforts differ from those of the search engines in Arriba Soft and Perfect 10 because, in those cases, the content was already in a digitized format online, and the search engines were merely acting as aggregators that indexed and displayed the content. As the court in Perfect 10 noted, in that case Google had not technically displayed or distributed Perfect 10’s photographs under the confines of the copyright statute because Google simply communicated the HTML address where the works were located rather than copying the images and because there was no physical copy of the works on Google’s servers for which Google could distribute.
The rights holders also would argue that Google went well beyond indexing and displaying by actively converting works into a new medium without permission, and this creates a presumption that the use is not fair. Moreover, despite Google’s claimed good intentions, Google stands to generate millions of dollars in advertising revenue through its unauthorized activities. Overall, this factor tilts in favor of the rights holders.
Second Factor: Nature of the Copyrighted Work
The focus of the second fair use factor is whether the infringed work was published or unpublished and whether the work was factual or creative. Fair use is a defense that is more likely available when the infringed work was previously published or is comprised of more factual content than a work that has not yet been published or whether the work contains more creative content. In the five cases discussed herein, this factor was not determinative to any of the court’s ultimate decision.
In both Arriba Soft and Perfect 10, the creative nature of the plaintiff’s photographs was offset by the fact that the images were already published online.48 In MP3.com and Napster the content was deemed highly creative, which hurt the fair use defense, and in Kirkwood the court held that the broadcast of a compilation of another’s copyrighted works (e.g., musical recordings, radio advertisements) was “unique and creative” rather than informational.
To date, there has been no analysis of the nature of the books Google scanned. So without knowing whether these books were predominantly factual or creative, it is difficult to speculate as to which side would come out ahead with respect to this fair use factor. However, assuming that there were more scanned books falling in the creative category than the factual category, then this prong would likely favor the rights holders. Nevertheless, Google would emphasize the fact that the scanned books were all previously published since they were scanned at public libraries. Because of the divided nature of this factor, neither side is likely to come out ahead here.
Third Factor: Amount and Substantiality of the Portion Used
Generally speaking, fair use is less likely to be found with higher amounts of copying. However, as evident by the Arriba Soft and Perfect 10 cases, entire works may be used in certain circumstances if the use of those works is highly transformative. The key determinant is whether the amount of copying was limited to what was necessary to carry out the intended purpose. Sometimes, a small amount can negate the fair use defense if that portion used was central to the work.
In both Arriba Soft and Perfect 10, the Ninth Circuit determined that it was necessary for the defendants to copy the entire image in order for it to be recognizable to users conducting the search and because of the purpose served by the defendant’s transformative use—specifically, providing access to information. Not surprisingly, in MP3.com, Napster, and Kirkwood, the fact that the defendants copied the entire works hurt their fair use claim because in each of those cases the court already determined that the infringing use was not transformative.
Focusing on the Arriba Soft and Perfect 10 decisions, Google would argue that it was reasonable and necessary to copy entire works in order to serve its intended transformative purpose of creating an accurate and comprehensive online book index. Also, while Google scanned entire books, it limited its display of those books to snippets.
The rights holders would argue that Google’s actions were no different than those of the defendants in MP3.com, Napster, and Kirkwood where wholesale copying was not allowed because the infringed works had been distributed in a new medium without any change in the character of the use. They also would argue that Google’s infringement extends beyond scanning to include permanent storage of each digitized book on Google’s servers, and an unauthorized scanning project of this magnitude should be rejected as a matter of public policy. In short, this factor arguably favors the rights holders.
Fourth Factor: Effect upon the Potential Market for the Work
The fourth fair use factor relates to whether the unauthorized use would impair the copyright owner’s marketability of the work. This factor has been considered the most important. Under this factor, courts seek answers to questions such as whether the copy of the work was obtained lawfully, how many copies of the copyrighted work are available on the market, and whether the copyright owner has a licensing system available for which it could exploit the work.
The impact that the unauthorized use had on the copyright owner’s marketability of the infringed works was a determining factor in the courts’ decisions. In Arriba Soft the court determined that there was no adverse affect on the market for plaintiff’s photographs because the miniature size of the thumbnails and their low resolution were not a substitute for the original works. The Perfect 10 court agreed with respect to full-size images but determined that defendant’s display of thumbnail images harmed plaintiff’s market for cell phone downloads. In MP3.com there was a potential adverse impact on the plaintiffs that the defendants could not overcome, and in Napster the negative effect was clear because defendant’s peer-to-peer sharing system reduced plaintiff’s CD sales and raised barriers to plaintiff’s entry into the digital downloading market. Finally, the court in Kirkwood said that defendant’s phone retransmission service sought to replace plaintiff as the supplier of those broadcasts, which, in the court’s opinion, was the kind of harm that copyright law aims to prevent.
Google would argue that the snippets that it displays do not adequately substitute for the book as a whole, and therefore, the snippets should not create an identifiable adverse impact on the marketability of the scanned books. Google may even claim that by providing online access to information about the digitized books, the Google Library Project may actually enhance the marketability of copyrighted books by generating increased awareness and interest. The rights holders would counter Google’s points by arguing that the sheer magnitude of the Google Library Project and resulting online book database is so large that it deters rights holders from seeking their own licensing agreements (which may or may not include the authorized online display of entire works) with other online providers. Weighing the two sides, this factor slightly favors Google.
An analysis of the four fair use factors in relation to the relevant case law demonstrates that the scales of equity tip towards the rights holders (i.e., factors one and three favor the rights holders, factor two is indeterminate, and factor four arguably tilts towards Google). In addition to these factors, it is also likely that the U.S. District Court for the Southern District of New York would factor in the massive scope and systematic nature of Google’s unauthorized copying, thereby denying Google’s fair use claim. The court would recognize that a ruling in Google’s favor would send the wrong message—a message that would effectively permit widespread copying of copyrighted works without authorization—as long as a convincing public benefit is emphasized by the copier. We can only hope that the court would go down the right path and preserve the exclusive statutory rights afforded the copyright owners.