The Authors Guild v. Google, Inc.

Filing 55

AMICUS BRIEF, on behalf of Amicus Curiae American Library Association, Association of College and Research Libraries and Association of Research Libraries, FILED. Service date 11/16/2012 by CM/ECF. [771350] [12-3200]--[Edited 11/19/2012 by KG] [Entered: 11/16/2012 03:02 PM]

12-3200-CV
IN THE

United States Court of Appeals
FOR THE SECOND CIRCUIT

THE AUTHORS GUILD, INC., Associational Plaintiff,
BETTY MILES, JOSEPH GOULDEN, and JIM BOUTON, individually
and on behalf of all others similarly situated,
Plaintiffs-Appellees,
v.

GOOGLE, INC.,
Defendant-Appellant.

On Appeal from the United States District Court
for the Southern District of New York (New York City)

BRIEF FOR AMICI CURIAE
THE AMERICAN LIBRARY ASSOCIATION,
THE ASSOCIATION OF COLLEGE AND
RESEARCH LIBRARIES, AND THE ASSOCIATION
OF RESEARCH LIBRARIES IN SUPPORT OF
DEFENDANT-APPELLANT AND REVERSAL

JONATHAN BAND
JONATHAN BAND PLLC
21 Dupont Circle NW, 8th Floor
Washington, DC 20036
[phone]

STEFAN M. MENTZER
WHITE & CASE LLP
1155 Avenue of the Americas
New York, NY 10036
[phone]

CORPORATE DISCLOSURE STATEMENT
Amici curiae the American Library Association, the Association of College
and Research Libraries, and the Association of Research Libraries state that none
of these entities has a parent corporation and that no publicly held corporation
owns 10% or more of their respective stock.

TABLE OF CONTENTS
INTEREST OF AMICI CURIAE ...............................................................................1
INTRODUCTION AND SUMMARY OF ARGUMENT ........................................2
ARGUMENT .............................................................................................................5
I. 

CLASS CERTIFICATION IN THIS CASE THREATENS TO
UNDERMINE THE PUBLIC INTEREST .....................................................5 
A.  Google Book Search Benefits Researchers, Scholars,
Librarians, Authors, and the Public at Large ............................................. 5 
B.  The Potentially Immense Damages Available in a Class Action
Threaten the Public Interest in Google Book Search ................................ 7 
C.  Class Certification Would Have a Chilling Effect on
Future Mass Digitization Efforts ............................................................... 9 

II.  THIS CASE PRESENTS NO COMMON ISSUE OF LAW OR FACT ........9 
A.  Common Issues Must Be Capable of Generating Answers
that Drive the Resolution of the Litigation .............................................. 10 
B.  There Are No Common Issues under Wal-Mart...................................... 11 
1.  Fair Use Is Not a Common Issue under Rule 23(a)(2) ......................11 
2.  Whether Google Is Liable Fails to Satisfy the
Commonality Requirement ................................................................14 
3.  There Are No Other Common Questions of Law or Fact .................15 
III.  USING SUBCLASSES TO DETERMINE FAIR USE IS AN
UNWORKABLE SOLUTION TO THE PROBLEMS INHERENT IN
LITIGATING THIS DISPUTE AS A CLASS ACTION .............................17 
IV.  CLASS CERTIFICATION WOULD MAKE IT VIRTUALLY
IMPOSSIBLE TO CRAFT AN INJUNCTIVE REMEDY ..........................22 
CONCLUSION ........................................................................................................25

i

TABLE OF AUTHORITIES

CASES
Authors Guild v. Google, Inc., 282 F.R.D. 384 (S.D.N.Y. 2012) ...................... passim
Authors Guild, Inc. v. HathiTrust, No. 11 CV 6351,
2012 WL 4808939 (S.D.N.Y. Oct. 10, 2012) ............................................. 8, 14, 22
Califano v. Yamasaki, 442 U.S. 682 (1979) .............................................................. 10
Cambridge Univ. Press v. Becker, 863 F. Supp. 2d 1190 (N.D. Ga. 2012) .. 19, 22, 24
Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569 (1994) .............................. 4, 13, 22
Chicago Bd. of Educ. v. Substance, Inc., 354 F.3d 624 (7th Cir. 2003).................... 23
Eldred v. Ashcroft, 537 U.S. 186 (2002) ................................................................... 11
FPX, LLC v. Google, Inc., 276 F.R.D. 543 (E.D. Tex. 2011) ............................. 13, 14
Maxtone-Graham v. Burtchaell, 631 F. Supp. 1432 (S.D.N.Y. 1986) ...................... 18
Wal-Mart Stores, Inc. v. Dukes, 131 S. Ct. 2541 (2011) .................................... passim
STATUTES, RULES, AND LEGISLATIVE MATERIALS
17 U.S.C. § 107 .......................................................................................................... 12
17 U.S.C. § 410 .......................................................................................................... 23
17 U.S.C. § 412 ............................................................................................................ 7
17 U.S.C. § 501 .................................................................................................... 15, 23
17 U.S.C. § 504 ........................................................................................................ 7, 8
Fed. R. Civ. P. 23 ................................................................................................ passim
H.R. Rep. No. 94-1476 (1976) ................................................................................... 23
S. Rep. No. 94-473 (1975) ......................................................................................... 18
ii

The American Library Association, the Association of College and Research
Libraries, and the Association of Research Libraries submit this amicus brief in
support of Google and request reversal of the District Court’s order certifying a
class of plaintiffs. All parties have consented to the submission of this brief.1
INTEREST OF AMICI CURIAE
The American Library Association (“ALA”), established in 1876, is a
nonprofit professional organization of more than 60,000 librarians, library trustees,
and other friends of libraries dedicated to providing and improving library services
and promoting the public interest in a free and open information society.
The Association of College and Research Libraries, the largest division of
the ALA, is a professional association of academic and research librarians and
other interested individuals. It is dedicated to enhancing the ability of academic
library and information professionals to serve the information needs of the higher
education community and to improve learning, teaching, and research.
The Association of Research Libraries (“ARL”) is a nonprofit organization
of 125 research libraries in North America. ARL’s members include university,
public, governmental, and national libraries. ARL programs and services promote

1

In accordance with Local Rule 29.1, no party’s counsel authored this brief in
whole or in part, no party and no party’s counsel contributed money that was
intended to fund preparing or submitting the brief, and no person other than amici
curiae and their counsel contributed money that was intended to fund preparing or
submitting the brief.
1

equitable access to and effective use of recorded knowledge in support of teaching
and research.
Collectively, amici represent more than 100,000 libraries and 350,000
librarians and other personnel in the United States that serve the needs of the
public in the digital age. The associations share a strong interest in the balanced
application of copyright law to new technologies and online services like Google
Book Search (“GBS”) that disseminate information, knowledge, and learning. In
particular, the library association amici have an interest in preserving the public
benefits that GBS offers and in ensuring the correct application and continuing
vitality of the fair use doctrine, including in the context of class action litigation.
INTRODUCTION AND SUMMARY OF ARGUMENT
The decision below should be reversed.
1. Google Book Search has become an essential research tool that benefits
scholars, students, and libraries, as well as authors themselves. Class certification,
however, would jeopardize the public’s continued access to GBS. Appellees seek
an award of the statutory minimum of $750 for each work used, and with millions
of copyrighted works in dispute, the damages could be immense. Such an award
would have no connection to the alleged injury suffered or the many benefits that
GBS provides to the public. Class certification has the potential to subject GBS to

2

potentially crippling damages and thus to eliminate or impair the public’s access to
and use of the service.
Class certification would also cause wider harm. It would set a precedent
that any mass digitization initiative is subject to class action claims and the
enormous damages exposure that accompanies them. The mere threat of having to
defend against such claims creates a disincentive for libraries, universities, and
others to engage in socially beneficial digitization efforts, even where there is a
strong fair use justification. Class certification thus could unintentionally stifle
further progress in digitizing and disseminating information and knowledge for the
benefit of the public.
2. To protect litigants and preserve the public interest, the Federal Rules
impose several requirements for class certification, including that there be issues of
law or fact common to the class. See Fed. R. Civ. P. 23(a)(2). An issue is
common only if it is capable of classwide resolution – in other words, where it has
the capacity to generate common answers apt to drive the resolution of the
litigation. Wal-Mart Stores, Inc. v. Dukes, 131 S. Ct. 2541, 2551 (2011).
There are no common issues that are capable of classwide resolution under
the standard the Supreme Court established in Wal-Mart. The two issues the
District Court found to be common – fair use and infringement – do not satisfy
Rule 23(a)(2). Fair use certainly is at the core of this dispute. But three of the four
3

statutorily-mandated fair use factors involve facts unique to each copyrighted work
and its usage. And none of the four factors may be examined in isolation; rather,
the inquiry must explore and weigh all of the factors together. Campbell v. AcuffRose Music, Inc., 510 U.S. 569, 578 (1994). The analysis is not necessarily the
same for all members of the class.
The only other issue identified by the District Court – infringement – is not a
common question; it is the ultimate question. An issue at such a high level of
generality does not satisfy Rule 23(a)(2), as Wal-Mart noted. If it did,
commonality would exist in every circumstance. There are no other common
issues.
3. The Court cannot overcome the lack of commonality, or predominance
for that matter, by employing subclasses to conduct its fair use analysis. The
District Court never explained how it would use subclasses. A closer look reveals
that using subclasses, even if appropriate as a legal matter, would require hundreds
and more likely thousands of fair use mini-trials. Subclasses cannot repair the
inherent flaws of litigating this dispute as a class action.
4. Class certification would make it virtually impossible for the District
Court to craft an injunctive remedy if liability is found. Because ownership of a
valid copyright is an element of a copyright infringement claim, the District Court
could not issue an injunction barring GBS from using works owned by non-class
4

members. It would have to decide who owns each work at issue, but definitive
copyright ownership information is not publicly available. Discovery of all class
members would be necessary, and the court or a jury would be burdened with
assessing the ownership status of millions of works – a nearly impossible task.
ARGUMENT
I.

CLASS CERTIFICATION IN THIS CASE THREATENS TO
UNDERMINE THE PUBLIC INTEREST
A.

Google Book Search Benefits Researchers, Scholars, Librarians,
Authors, and the Public at Large

From the perspective of the amici library associations, Google Book Search
advances the public interest in a number of important ways.
GBS has become an essential research tool for scholars, students, and
libraries. By letting users search for and identify relevant works, GBS allows
researchers to get started on a path to access these materials through the physical
holdings of a library or through other means like interlibrary loan. To librarians,
GBS is indispensable for identifying valuable research sources. GBS in particular
benefits students and faculty at small and remote educational institutions – many of
which have limited resources and face significant challenges in identifying and
accessing materials. Where a library does not hold a particular book in its
collection, GBS helps librarians to decide whether to request a book by interlibrary
loan, conserving strapped library budgets and speeding research. At
geographically isolated institutions, getting books through interlibrary loan can
5

take weeks, with significant postage costs. By using GBS to identify the specific
materials needed, a library can narrow its request and obtain the material in a day
or two. And GBS has become a crucial tool for finding and checking citations,
saving libraries and scholars significant time and allowing them to focus their
efforts more productively elsewhere.
GBS is valuable not just for research, but also for developing and making
the most of existing library collections. In amici’s experience, GBS has led
librarians to identify resources that they purchase for their libraries, and it has
assisted faculty members to order books for their own private collections. GBS
also helps to identify material that is housed by a library itself but that otherwise
would be difficult if not impossible to find.
For all these reasons, GBS has become an integral part of this nation’s
education system. GBS enhances scholarship and opens a universe of knowledge
and learning that otherwise would be off limits for many.
GBS also benefits authors, despite Appellees’ claims to the contrary. Most
authors appear to approve of GBS. See A244; Br. of Amici Curiae Academic
Authors in Supp. of Def.-Appellants and Reversal. The service helps readers find
authors, which grows their audience and, not incidentally, improves sales. GBS is
also valuable to authors as a research tool. GBS allows authors to tell whether a

6

book will assist their scholarship and writings, and it helps authors decide whether
to borrow or buy books that are not in their own collections.
B.

The Potentially Immense Damages Available in a Class Action
Threaten the Public Interest in Google Book Search

Class certification threatens to deprive the public’s access to GBS and the
many benefits it provides, by potentially giving rise to statutory damages that will
disable or destroy the service. The Copyright Act allows, under certain
circumstances, a plaintiff in a copyright infringement action to recover statutory
damages, ordinarily in the range of $750 to $30,000 per work infringed. 17 U.S.C.
§ 504(c). Every member of Appellees’ proposed class is eligible to elect statutory
damages. See 17 U.S.C. § 412; SPA1 (“‘Books’ means each full-length book
published in the United States in the English language and registered with the
United States Copyright Office within three months after its first publication.”).
Appellees seek an award of the $750 statutory minimum for each work owned by a
class member. Mem. of Law in Supp. of Pl.’s Mot. for Partial Summ. J. 47, ECF
No. 1050.
Class certification is especially problematic in this case because of the
millions of works in suit. If GBS is found to infringe, the total statutory damages
award will not take into account the immense benefits that GBS provides to the
public. In a case where fair use is at the core of the dispute and a careful
examination and weighing of numerous factors is essential for a determination of
7

liability, the District Court will have no discretion to fashion a monetary remedy
below the statutory minimum that reflects the circumstances of the use and the
impact on the public. Nor does the Copyright Act require Appellees to prove that
$750 per work fairly compensates class members for their purported injuries.
Indeed, it is unlikely that Appellees ever will try to make such a showing, other
than to offer anecdotal evidence and conclusory arguments.2
Rather, class certification would create a toxic combination of fixed
damages and large numbers of works that potentially exposes GBS to an award in
the billions of dollars.3 The amici library associations are confident that Google
ultimately will prevail on its fair use defense in this case, just as HathiTrust did on
summary judgment with The Authors Guild and other individual plaintiffs
concerning the Google-assisted mass digitization project. See Authors Guild, Inc.
v. HathiTrust, No. 11 CV 6351, 2012 WL 4808939 (S.D.N.Y. Oct. 10, 2012). But
there is a real concern that infringement liability, however remote a possibility,
will give rise to a massive damage award that will harm the public interest. Such
an award could force the shutdown of parts or all of GBS, significantly limit the

2

Appellees’ motion for summary judgment submitted to the District Court made
no attempt to quantify the market harm suffered under the fourth fair use factor, or
to explain how or why $750 per book appropriately compensates their supposed
losses. See Mem. of Law in Supp. of Pl.’s Mot. for Partial Summ. J. 41-47, ECF
No. 1050.
3
Exposure still reaches into the billions of dollars if Google is found to be an
innocent infringer and the court reduces statutory damages for works in suit to
$200. See 17 U.S.C. § 504(c)(2).
8

works available in the database and thereby reduce its usefulness, or restrict access
to members of the public including researchers, scholars, students, and libraries
who depend on GBS. None of these outcomes would benefit the public. In a case
where a majority of authors appear to approve of GBS, it is manifestly unfair to
give the few representative plaintiffs here such leverage.
C.

Class Certification Would Have a Chilling Effect on
Future Mass Digitization Efforts

Class certification has implications beyond this case. It would create a
precedent for certifying classes of plaintiffs in other similar situations. That
particularly affects libraries and universities that may be contemplating large-scale
digitization projects. These institutions lack the resources to litigate complex class
action lawsuits and defend against massive damages claims, and they will be
disinclined to engage in socially beneficial digitization efforts even where they
have strong fair use justifications. Class certification would, in essence put the
brakes on future mass digitization initiatives and technological innovations, no
matter how much the public may benefit from them.
II.

THIS CASE PRESENTS NO COMMON ISSUE OF LAW OR FACT
Precisely because class actions have potentially far-reaching consequences

for litigants and the public interest, the Federal Rules of Civil Procedure limit the
ability of parties to obtain class status. One of these limitations is that there must

9

be issues of law or fact common to the class. Fed. R. Civ. P. 23(a)(2). No
common issues are presented here.
A.

Common Issues Must Be Capable of Generating
Answers that Drive the Resolution of the Litigation

Class actions are an exception to the usual rule that litigation is conducted by
and on behalf of the individual named parties only. Califano v. Yamasaki, 442
U.S. 682, 700-01 (1979). To justify a departure from that rule, the Federal Rules
of Civil Procedure require a plaintiff to establish four requirements before a class
can be certified. Fed. R. Civ. P. 23(a). Our brief focuses on one of these
requirements – commonality. See id. 23(a)(2) (allowing a member of a class to sue
as a representative party on behalf of all members only if there are “questions of
law or fact common to the class”).
Until recently, the commonality requirement was easily satisfied. That
changed a little more than a year ago with Wal-Mart Stores v. Dukes. In WalMart, the Supreme Court reversed the certification of a nationwide class of 1.5
million female employees against whom Wal-Mart allegedly discriminated in
violation of Title VII of the Civil Rights Act of 1964. The Court held that there
was no question of law or fact common to the class, and in so ruling, articulated a
more stringent commonality standard. The commonality requirement, the Court
noted, is easy to misread, as any competently crafted class complaint literally
raises common questions. Wal-Mart Stores, 131 S. Ct. at 2551. The lawsuit must
10

depend on a common contention that is capable of classwide resolution, meaning
that determination of its truth or falsity will resolve an issue that is “central to the
validity of each one of the claims in one stroke.” Id. The Court concluded that
what matters is not the ability of the plaintiff to raise common questions, even in
droves, but rather “the capacity of a classwide proceeding to generate common
answers apt to drive the resolution of the litigation.” Id. (quoting Richard
Nagareda, Class Certification in the Age of Aggregate Proof, 84 N.Y.U. L. Rev.
97, 131-32 (2009)). That is the standard that governs here.
B.

There Are No Common Issues under Wal-Mart

The District Court found that this case presents two common questions:
whether Google’s actions constitute an infringement of the class members’
copyright interests, and whether Google’s use of snippets of these works
constitutes fair use. Authors Guild v. Google, Inc., 282 F.R.D. 384, 394 (S.D.N.Y.
2012). Neither is common to the class.
1.

Fair Use Is Not a Common Issue under Rule 23(a)(2)

Fair use, the District Court noted, is Google’s principal defense in this
litigation. Id. at 387. Fair use is one of the First Amendment safeguards built into
the Copyright Act, Eldred v. Ashcroft, 537 U.S. 186, 219-20 (2002), and there is a
strong policy interest in maintaining the doctrine’s vitality across a variety of

11

factual situations. Congress accordingly wrote Section 107 of the Copyright Act to
require a particularized inquiry into the facts of each work and its usage.
A court determines whether the use made “of a work in any particular case”
is fair. 17 U.S.C. § 107. It does so by considering four (and other, if appropriate)
factors. Three of these factors involve facts unique to the work and the usage. The
second fair use factor examines the nature of each copyrighted work in suit, such
as whether the work is creative or informational, or in-print or out-of-print. The
third factor considers the relative amount and substantiality of the copyrighted
work used. The fourth factor considers the economic effect on the market for or
value of the copyrighted work, for example whether the use negatively impacts or
benefits the economic prospects for the work – again, an inquiry into facts
particular to the work. Google’s opening brief details the ways in which the
second, third, and fourth factors present predominant individual issues, and those
arguments need not be repeated here. See Br. for Appellant 29-35. Suffice it to
say, none of the three factors present common issues among class members.
Google has acknowledged that the first fair use factor – the purpose and
character of the use – could be resolved on a classwide basis. Def.’s Opp’n to Mot.
for Class Certification 18, ECF No. 1000. The first factor does not satisfy the WalMart commonality standard, however. The Copyright Act requires all four factors
to be considered together in determining whether the use made of a work in any
12

particular case is fair. No single fair use factor may be considered in a vacuum. A
factor must be considered and weighed together with the other factors in the same
analysis. Campbell, 501 U.S. at 578 (“Nor may the four statutory factors be
treated in isolation, one from another. All are to be explored, and the results
weighed together, in light of the purposes of copyright.”). A single fair use factor
viewed in isolation, even if it could be decided on a classwide basis, has no
independent significance and will not generate common answers to the question of
fair use.
A recent district court decision is instructive on how a multifactor analysis
that contains factors capable of classwide resolution nevertheless fails to satisfy the
commonality requirement of Rule 23(a)(2). In FPX v. Google, a trademark
infringement lawsuit against Google, the court found no common issues and
denied plaintiffs’ motion for class certification. FPX, LLC v. Google, Inc., 276
F.R.D. 543, 549-50 (E.D. Tex. 2011). There, the plaintiffs alleged that Google
infringed their trademarks by selling keywords containing their marks to third
parties, so that when an internet user searches for the trademark owner’s mark
using Google’s search engine, a competitor’s advertisement appears next to the
search results. The plaintiffs’ burden was to prove that Google’s use is likely to
cause initial interest confusion, based on a weighing of seven nonexclusive factors.

13

Two of those factors conceivably could be decided on a classwide basis.4 The
district court concluded that, even if Google’s policy resulted in initial interest
confusion with regard to FPX’s trademark, that did not necessarily mean Google’s
policy would result in confusion with regard to the other class members’ marks.
Id. at 550. For that reason, plaintiffs’ common contention that Google’s policy of
selling trademarks as keywords leads to initial interest confusion was not capable
of classwide resolution and did not satisfy the commonality requirement. Id. The
same reasoning should apply here. Even though the purpose and character of use
factor can be resolved classwide, the fair use analysis is not necessarily the same
for all class members.5
2.

Whether Google Is Liable Fails to Satisfy the Commonality
Requirement

The one other common issue identified by the District Court – whether
Google’s actions infringe the class members’ copyrights – merely states the

4

The likelihood of confusion analysis considers (1) the type of trademark allegedly
infringed, (2) the similarity between the two marks, (3) the similarity of the
products or services, (4) the identity of the retail outlets and purchasers, (5) the
identity of the advertising media used, (6) the defendant’s intent, and (7) any
evidence of actual confusion. Id. at 549. Factors 5 and 6 are potentially capable of
classwide resolution.
5
Of course, if the District Court were to find that the entire GBS project is
legitimate based on the transformative purpose and character of the use, then it
could find fair use without having to consider each individual work. See, e.g.,
HathiTrust, 2012 WL 4808939, at *8-14. But if, as Google notes in its opening
brief, the District Court were to reject the contention that the entire GBS project is
fair use, Google is entitled to present, and the Court must consider, a fair use
defense as to uses of individual works. Br. for Appellant 2.
14

ultimate question of whether the defendant is liable as a matter of law. Under the
Copyright Act, a defendant is liable if it infringed the plaintiff’s copyrights. 17
U.S.C. § 501(a). Defining the issue at this highest level of generality renders the
common issue requirement meaningless. Commonality would always be satisfied
because in every case the issue is whether the defendant is liable. The Supreme
Court recognized that such formulations are not sufficient when it ruled that
commonality does not mean merely that the class members all suffered a violation
of the same provision of law. Wal-Mart Stores, 131 S. Ct. at 2551.
3.

There Are No Other Common Questions of Law or Fact

In their opening brief with the District Court requesting class certification,
Appellees listed what they saw as the common issues in the case. In addition to
fair use and whether Google’s actions constitute copyright infringement, Appellees
suggested as common issues: (1) “[w]hether Google digitized Books, distributed
copies of Books to libraries, and publicly displayed content from Books on its
commercial website without authorization from the copyright owners as part of its
Library Project”; (2) “[w]hether Google’s actions are ongoing, such that injunctive
relief is appropriate”; and (3) “[w]hether members of the Class are entitled to

15

statutory damages.” 6 Mem. of Law in Supp. of Pl.’s Mot. for Class Certification
10, ECF No. 990. None satisfies Rule 23(a)(2).
The first of these questions – what actions Google took – is not in dispute.
Google digitized copyrighted works, gave copies to libraries, and publicly
displayed snippets online. Undisputed facts like these do not generate common
answers that are apt to drive the resolution of the litigation. See Wal-Mart Stores,
131 S. Ct. at 2551.
The second and third questions go to the remedies available if liability is
found, not the merits of the underlying claims. Again, these do not generate
common answers that drive resolution of the case. See id.
All three questions are of the same type that the Supreme Court rejected as
not satisfying commonality in Wal-Mart: “Do all of us plaintiffs indeed work for
Wal-Mart? Do our managers have discretion over pay? Is that an unlawful
employment practice? What remedies should we get?” Id. While certainly these
are questions, none of them are questions of law or fact that are relevant to the

6

Appellees slightly recast these issues in their reply brief to the District Court: (1)
“[d]oes Google scan in-copyright books in their entirety in its Library Project,
distribute entire digital copies of such books to libraries, and/or display verbatim
expression from such books on its search website, all without copyright owner
permission?”; (2) “[s]hould statutory damages of $750 per book be assessed?; and
(3) “[w]hat injunctive and equitable relief is appropriate to address Google’s
conduct?” Reply Mem. of Law in Supp. of Pl.’s Mot. for Class Certification 13,
ECF No. 1008.
16

commonality inquiry, and merely reciting them is not sufficient to obtain class
certification. See id.
III.

USING SUBCLASSES TO DETERMINE FAIR USE IS AN
UNWORKABLE SOLUTION TO THE PROBLEMS INHERENT IN
LITIGATING THIS DISPUTE AS A CLASS ACTION
In the portion of its opinion addressing the predominance requirement of

Rule 23(b)(3), the District Court noted that the question of fair use could be
determined on a “sub-class-wide basis.” In particular, “[t]he court would
determine whether the defense applies to a particular type of book, obviating the
need to evaluate each book individually.” Google, 282 F.R.D. at 395. The District
Court’s suggestion is telling: it indicates there are issues that are not capable of
classwide resolution and that require individualized attention. Regardless, creating
subclasses to decide fair use is not feasible, and it would not resolve the lack of
commonality under Rule 23(a)(2) or predominance under Rule 23(b)(3).
The District Court did not explain how it would employ subclasses in its fair
use analysis. Even if using subclasses were appropriate, the Court would have to
construct a multiplicity of groupings to account for the different types of works,
the relative amounts used, and the economic effects of the usages. Take, for
example, the second fair use factor. The District Court found that it could create
subclasses of owners of fiction, nonfiction, poetry, cookbooks, photographs, and
illustrations. Google, 282 F.R.D. at 390. More groupings would be necessary,

17

because the second factor involves additional considerations like whether works
are in print or out of print. See Maxtone-Graham v. Burtchaell, 631 F. Supp. 1432,
1438 (S.D.N.Y. 1986); S. Rep. No. 94-473, at 64 (1975) (“whether or not the work
is available to the potential user” is a “key, though not necessarily determinative,
factor in fair use.”). Thus, the District Court would have to create at least twelve
subclasses (i.e., one for in-print works of fiction, one for out-of-print works of
fiction, one for in-print works of nonfiction, etc.).
That number is conservative. Distinctions can be drawn between books
published inside and outside the U.S. – the latter being more amenable to fair use
because of their unavailability in this country. And there are many more genres of
works than just the six the District Court identified – perhaps dozens – each
requiring its own distinct analysis under the second factor. All works of nonfiction
cannot be lumped together into one group as the District Court suggested, given
the enormous variation in content, audience, and pricing. Nonfiction trade books,
such as popular biographies and histories, differ markedly in nature from
professional works like legal, medical, and engineering writings, which in turn are
distinct from academic works like textbooks and monographs. Likewise, works of
fiction cannot be fairly considered in one subclass given their enormous literary
range – romance, mystery, humor, historical fiction, science fiction, children’s and
young adult literature, to name a few.
18

The third and fourth factors would require subclasses as well – a point that
the District Court neglected to address. The third fair use factor considers the
relative amount of the portion used (a quantitative measure) and the substantiality
of the portion used (a qualitative measure). The Court would have to construct at
least nine subgroups to analyze whether the display of book snippets is fair use:
one subgroup for works where a small quantitative amount and a small qualitative
amount was displayed, another for works with a small quantitative and
intermediate qualitative taking, another for works with a small quantitative and
large qualitative taking, and so on. These nine subgroups are not the end of the
story. Given the tens of millions of volumes scanned into Google’s database, it is
conceivable that many works were never, or rarely, accessed by or displayed to the
public. In such circumstances, the mere reproduction of copies into the database is
de minimis and not subject to copyright liability, or at the very least weighs the
third factor strongly in favor of fair use. See Cambridge Univ. Press v. Becker,
863 F. Supp. 2d 1190, 1245-46, 1248, 1251, 1252, 1253, 1265, 1298-99, 1314,
1337 (N.D. Ga. 2012) (finding infringement of copyrighted works de minimis and
not giving rise to liability when digital copies uploaded to a university electronic
reserves system were accessed only a few times, likely by a staff librarian,
professors, or counsel for the parties). Thus, additional subgroups are required for

19

works whose snippets GBS users viewed frequently, moderately, infrequently, or
not at all.
The fourth fair use factor would require multiple subclasses as well.
Whether a work suffered economic harm or enjoyed economic benefits by
inclusion in GBS turns on a host of unique factual considerations, including the
availability of the work, its prior licensing (if any) in the marketplace, the
economic interests of the author, and the channels of trade in which it typically is
sold. For example, trade, professional, and educational and academic books are
sold in different channels, each with its own economic characteristics relevant to a
market harm analysis. It is not clear that any arrangement of subclasses could
account for all of the necessary considerations. But even assuming subclasses
could accommodate analysis of the fourth factor, at a minimum three subclasses
would be required: one for works that suffer economic harm from the use, one for
works that benefit economically from the use, and one for works where the harms
and benefits are neutral. These broad categories, of course, ignore facts crucial to
the analysis.
After devising subclasses, the District Court would then have to combine
them, factor by factor, in order to assess fair use. Using just the limited number of

20

subclasses given above7 yields 324 separate combinations. In effect, the court or a
jury tasked with determining fair use would have to conduct at least 324 distinct
fair use mini-trials. That number will be much higher, likely in the thousands,
when the additional necessary subgroups discussed above are taken into account.
The large number of subclasses and subclass combinations call into doubt
the District Court’s conclusion that there is no need to evaluate each individual
work. To conduct the analysis by subclass, each of the millions of copyrighted
works in dispute would have to be identified and examined – first, to determine if
they are owned by class members and subject to the lawsuit, and for those that are,
then to slot each work, one-by-one into the subgroup that best describes it. The
District Court failed to recognize this when it concluded, without analysis, that it
could determine whether the defense applies to a particular type of book without
the need to evaluate each book individually. Google, 282 F.R.D. at 395. Each
book would have to be evaluated, by the court or a jury, in a painstaking, timeconsuming, and likely error-prone and dispute-ridden process.
The hundreds, if not thousands, of separate fair use analyses required in a
class action of this magnitude, and the necessity of examining each work in suit,
confirm that fair use is not a common issue capable of classwide resolution and

7

That is, twelve subclasses for factor two, nine subclasses for factor three, and
three subclasses for factor four.
21

that Appellees cannot satisfy the predominance requirement.8 Indeed, use of
subclasses may be wholly inappropriate because it runs contrary to the wellestablished method of considering fair use on a case-by-case basis. See Campbell,
501 U.S. at 578. The better approach is to revert to the traditional rule applicable
to civil litigation in this country and embodied in Rule 23(a) – requiring an
individual named party that believes it was injured to assert claims and prove them
in court.
IV.

CLASS CERTIFICATION WOULD MAKE IT VIRTUALLY
IMPOSSIBLE TO CRAFT AN INJUNCTIVE REMEDY
In addition to damages, Appellees seek injunctive relief. If the class is

certified and GBS found infringing, implementing an injunction would be
incredibly burdensome if not impossible for the District Court.

8

This is not to say the fair use analysis can never be conducted when multiple
plaintiffs or works are at issue. Courts outside of the class action setting have
appropriately decided fair use in such circumstances. See, e.g., HathiTrust, 2012
WL 4808939, at *8-14; Cambridge Univ. Press, 863 F. Supp. 2d 1190. The
HathiTrust and Cambridge University Press courts considered far fewer
copyrighted works than at issue here, and each assessed fair use with respect to the
works in its own case rather than the works at issue in millions of cases that a class
action effectively would combine. Class actions are not designed, as a matter of
law or practicality, to assess fair use on the vast scale as Appellees propose.
Moreover, HathiTrust and Cambridge University Press concluded that the
digitization projects at issue in those cases were entirely justified. In such
circumstances, a court need not consider each work allegedly infringed. But if the
District Court were to reject the contention that the entire GBS project is fair use,
the Court is required to consider a fair use defense for each work in dispute. See
Br. for Appellant 2. A class action is not suited for that undertaking.
22

Ownership of a valid copyright is an element of a copyright infringement
claim. 17 U.S.C. § 501(b). If liability were established, it would be unreasonable
and inequitable to bar GBS from using works that are owned by non-class
members, especially given the significant adverse impact on the public of such a
broad injunction. Any injunction would have to be limited to works that are owned
by class members. See Chicago Bd. of Educ. v. Substance, Inc., 354 F.3d 624, 632
(7th Cir. 2003) (vacating as overbroad an injunction that was not limited to the
secure tests owned by the plaintiff). But sorting out the works subject to an
injunction from those that are not would present immense challenges. Just as with
the fair use analysis, it would require examining each individual work – a difficult
if not impossible task given the millions of works at issue.
Contrary to the District Court’s suggestion, see Google, 282 F.R.D. at 390,
comprehensive copyright ownership information is not publicly available. While
timely-filed copyright registrations provide prima facie evidence of copyright
ownership, 17 U.S.C. § 410(c), they are not conclusive. The Copyright Office
does not independently verify the information submitted in a registration
application. See H.R. Rep. No. 94-1476, at 157 (1976) (“It is true that, unlike a
patent claim, a claim to copyright is not examined for basic validity before a
certificate is issued.”). Nor is there a requirement that transfers or licenses of
copyright be recorded with the Copyright Office, making a registration certificate,
23

at best, a snapshot of who owned the copyright at the moment the application was
filed. There is a significant risk that a registration certificate contains ownership
information that is inaccurate, out-of-date, or false, and for that reason the
Copyright Act entitles a defendant to rebut the presumption that attaches to a
timely-filed registration. Once a defendant effectively challenges the validity of
the copyright, the burden shifts back to the plaintiff to prove it. Id.
Google is entitled to defend itself, and it would be entitled to discovery of
each class member to determine if he or she owns a valid interest. Establishing
ownership will prove difficult in a class of this size. Copyright ownership may
have been transferred from a class member to a non-class member, or the copyright
may be owned by a non-class member as a work made for hire. Where evidence of
ownership is missing – as it invariably will be with such a large number of works –
it cannot be presumed. See Cambridge Univ. Press, 863 F. Supp. 2d at 1221-23
(district court refused to find that works were joint works co-owned by plaintiff
where there was no evidence of collaboration and no evidence of mutual intent to
create a joint work, and refused to presume that missing contracts existed or that
missing contracts would contain particular provisions assigning rights to the
plaintiff).
A class action would open the door to burdensome litigation into ownership
issues specific to each class member and would require the District Court to
24

adjudicate ownership disputes, as Google aptly puts it, in a series of book-by-book
mini-trials. Br. for Appellant 3. Just as with analyzing fair use, a class action of
this scale is not suited for crafting injunctive remedies.
CONCLUSION
Google Book Search is a valuable resource for researchers, scholars,
libraries, and authors, and it makes vast amounts of information and learning far
more accessible to the public than ever before possible. The public has a strong
interest in having continued access to GBS – an interest that class certification
endangers. Class certification is not appropriate as legal, practical, or policy
matter, and the decision of the District Court should be reversed.

November 16, 2012

Respectfully submitted,

JONATHAN BAND
JONATHAN BAND PLLC
21 Dupont Circle NW, 8th Floor
Washington, DC 20036
[phone]
[e-mail]

/s/ Stefan M. Mentzer
STEFAN M. MENTZER
WHITE & CASE LLP
1155 Avenue of the Americas
New York, NY 10036
[phone]
[e-mail]
Attorneys for Amici Curiae  
American Library Association,
Association of College and Research
Libraries, and Association of Research
Libraries

25

CERTIFICATE OF COMPLIANCE
I hereby certify that on this 16th day of November, 2012:
1.

This brief complies with the type-volume limitation of Federal Rule of

Appellate Procedure 32(a)(7)(B) because the brief contains 6,093 words, excluding
the parts of the brief exempted by Federal Rule of Appellate Procedure
32(a)(7)(B)(iii); and
2.

This brief complies with the typeface requirements of Federal Rule of

Appellate Procedure 32(a)(5) and the type style requirements of Federal Rule of
Appellate Procedure 32(a)(6) because the brief has been prepared in a
proportionally spaced typeface using Microsoft Word in 14-point Times New
Roman.

/s/ Stefan M. Mentzer
STEFAN M. MENTZER

CERTIFICATE OF ELECTRONIC FILING
I hereby certify that on this 16th day of November, 2012, I caused a PDF
version of the foregoing brief to be filed electronically using the CM/ECF system.
Prior to transmittal, the PDF was scanned for viruses and no viruses were detected.

/s/ Stefan M. Mentzer
STEFAN M. MENTZER

CERTIFICATE OF SERVICE
I hereby certify that on this 16th day of November, 2012, I caused the
foregoing brief to be filed electronically using the CM/ECF system, which will
send notification of such filing to counsel of record.

/s/ Stefan M. Mentzer
STEFAN M. MENTZER