The Authors Guild, Inc. et al v. Hathitrust et al

Filing 123

FIRST MEMORANDUM OF LAW in Support re: 100 MOTION for Summary Judgment.. Document filed by DIGITAL HUMANITIES AND LAW SCHOLARS. (Colman, Charles)

UNITED STATES DISTRICT COURT
SOUTHERN DISTRICT OF NEW YORK
------------------------------------------------------X
X
THE AUTHORS GUILD, INC. et al.,
X
X
Plaintiffs,
X
X
v.
X
X
HATHITRUST et al.,
X
X
Defendants.
X
X
------------------------------------------------------X

11 Civ. 06351 (HB)
ECF Case

BRIEF OF DIGITAL HUMANITIES AND LAW SCHOLARS
AS AMICI CURIAE IN PARTIAL SUPPORT OF
DEFENDANTS’ MOTION FOR SUMMARY JUDGMENT
CHARLES COLMAN LAW, PLLC
Charles Colman (CC-1133)
30 West 70th Street
New York, NY 10023
Phone: [phone]
Facsimile: 212-960-8969
E-mail: [e-mail]
Jason Schultz*
Assistant Clinical Professor of Law
UC Berkeley School of Law
[e-mail]
Matthew Sag*
Associate Professor
Loyola University of Chicago School of Law
[e-mail]
Counsel for Amici
* Filed in their individual capacity, and not on behalf of their institutions

TABLE OF CONTENTS

TABLE OF AUTHORITIES ............................................................................................. iii

STATEMENT OF INTEREST OF AMICI......................................................................... 1

SUMMARY OF ARGUMENT .......................................................................................... 1

ARGUMENT ...................................................................................................................... 3

I.
  The Freedom to Make Nonexpressive Use of Copyrighted Works is Vital to the
“Progress of Science” in the Digital Humanities ................................................................ 3

II.
  Copyright Law Does Not Protect Nonexpressive Aspects of Works........................ 11

A.
  The Idea/Expression Distinction ........................................................................... 11

B.
  Section 102(b) ....................................................................................................... 12

C.
  Merger and Scènes à Faire.................................................................................... 13

D.
  Fact/Expression Distinction .................................................................................. 13

E.
  Nonexpressive Metadata Does Not Implicate the Statutory Rights of the Copyright
Holder ........................................................................................................................... 14

F.
  Nonexpressive Metadata Is Also Noninfringing Because It Does Not Allow the
Public to Perceive the Expressive Content of a Work .................................................. 17

III.
  Text Mining Creates Value by Facilitating the Advancement of Our Collective
Knowledge; To Protect That Value, Mass Digitization and Similar Intermediate Copying
For Data Mining and Other Nonexpressive Purposes Should Be Considered "Fair Use" 19

A.
  Nonexpressive Copying to Expand Our Knowledge in the Digital Humanities Is
An Activity of the Sort that Copyright Law Should Favor, Through Fair Use ............ 19

B.
  The Nature of the Works in Question Is Neutral to the Fair Use Analysis of Mass
Digitization for the Advancement of Digital Humanities Research and Scholarship .. 22

C.
  To the Extent Relevant, Mass Digitization Uses a Reasonable “Amount and
Substantiality” of the Works in Question, in Light of the Socially Beneficial Purpose of
Facilitating Data Mining for the Advancement of the Digital Humanities .................. 23

D.
  Allowing Intermediate Copying in Order to Enable Nonexpressive Uses Does Not
Harm the Market for the Original Works in a Legally Cognizable Manner, As The
Practice Does Not Implicate the Works' Expressive Aspects in Any Way .................. 24

ii

TABLE OF AUTHORITIES
Cases
Arnstein v. Porter,
154 F.2d 464 (2d Cir. 1946)........................................................................................ 19
A.V. ex rel. Vanderhye v. iParadigms, LLC,
562 F.3d 630 (4th Cir. 2009) ...................................................................... 3, 22, 24, 25
Basic Books, Inc. v. Kinko's Graphics Corp.,
758 F. Supp. 1522 (S.D.N.Y. 1991)....................................................................... 23-24
Bill Graham Archives v. Dorling Kindersley Ltd.,
448 F.3d 605 (2d Cir. 2006).................................................................................. 21, 23
Bond v. Blum,
317 F.3d 385 (4th Cir. 2003) ................................................................................ 22, 25
Cambridge Univ. Press v. Becker,
No. 08 Civ. 01425 (N.D. Ga. May 11, 2012) ............................................................. 18
Campbell v. Acuff-Rose Music, Inc.,
510 U.S. 569 (1994) ...................................................................... 20, 21, 22, 23, 24, 25
Castle Rock Entm’t v. Carol Publishing Grp.,
150 F.3d 132 (2d Cir. 1998).................................................................................. 16, 17
Davis v. United Artists, Inc.,
547 F. Supp. 722 (S.D.N.Y. 1982).............................................................................. 19
Feist Publ’ns, Inc. v. Rural Tel. Serv. Co., Inc.,
499 U.S. 340 (1991) .............................................................................................. 13, 16
Fisher v. Dees,
794 F.2d 432 (9th Cir. 1986) ...................................................................................... 25
Fuld v. Nat’l Broad. Co., Inc.,
390 F. Supp. 877 (S.D.N.Y. 1975).............................................................................. 19
Golan v. Holder,
132 S. Ct. 873 (2012) .................................................................................................. 12
Harper & Row Publishers, Inc. v. Nation Enters.,
471 U.S. 539 (1985) ......................................................................................... 11-12, 23
iii

Hasbro Bradley, Inc. v. Sparkle Toys, Inc.,
780 F.2d 189 (2d Cir. 1985)........................................................................................ 17
Hoehling v. Universal City Studios, Inc.,
618 F.2d 972 (2d Cir. 1980).................................................................................. 13, 14
Kelly v. Arriba Soft Corp.,
336 F.3d 811 (9th Cir. 2002) ................................................................................ 20, 25
Kregos v. Associated Press,
937 F.2d 700 (2d Cir. 1991)........................................................................................ 13
Madrid v. Chronicle Books,
209 F. Supp. 2d 1227 (D. Wyo. 2002) ........................................................................ 19
Matthew Bender & Co., Inc. v. West Pub. Co.,
158 F.3d 674 (2nd Cir. 1998)...................................................................................... 18
MyWebGrocer, LLC v. Hometown Info, Inc.,
375 F.3d 190 (2d Cir. 2004)........................................................................................ 13
Nat’l Basketball Ass’n v. Motorola, Inc.,
105 F.3d 841 (2nd Cir. 1997)...................................................................................... 14
New Era Publ’ns Int’l, ApS v. Carol Pub. Grp.,
904 F.2d 152 (2d Cir. 1990)........................................................................................ 24
N.Y. Mercantile Exch., Inc. v. IntercontinentalExchange, Inc.,
497 F.3d 109 (2d Cir. 2007)........................................................................................ 13
N.Y. Times Co. v. Tasini,
533 U.S. 483 (2001) .................................................................................................... 18
NXIVM Corp. v. Ross Inst.,
364 F.3d 471 (2d Cir. 2004)........................................................................................ 21
Perfect 10, Inc. v. Amazon.com, Inc.,
508 F.3d 1146 (9th Cir. 2007) .................................................................... 3, 20, 22, 24
Peter F. Gaito Architecture, LLC v. Simone Dev. Corp.,
602 F.3d 57 (2d Cir. 2010).......................................................................................... 12
Religious Tech. Ctr. v. Lerma,
908 F. Supp. 1362 (E.D. Va. 1995) ............................................................................ 22
Reyher v. Children’s Television Workshop,
iv

533 F.2d 87 (2d Cir. 1976).......................................................................................... 12
Sega Enters. Ltd. v. Accolade, Inc.,
977 F.2d 1510 (9th Cir. 1992) .................................................................... 3, 22, 23, 25
Sony Computer Entm’t, Inc. v. Connectix Corp.,
203 F.3d 596 (9th Cir. 2000) ............................................................................ 3, 22, 24
Sony Corp. of Am. v. Universal City Studios, Inc.,
464 U.S. 417 (1984) .................................................................................................... 12
Stromback v. New Line Cinema,
384 F.3d 283 (6th Cir. 2004) ...................................................................................... 19
Tufenkian Imp./Exp. Ventures, Inc. v. Einstein Moomjy, Inc.,
338 F.3d 127 (2d Cir. 2003)........................................................................................ 12
Ty, Inc. v. Publ’ns Int’l Ltd.,
292 F.3d 512 (7th Cir. 2002) ...................................................................................... 17
Warner Bros. Entm’t Inc. v. RDR Books,
575 F. Supp. 2d 513 (S.D.N.Y. 2008)............................................................. 15, 16, 17
Walker v. Time Life Films, Inc.,
615 F. Supp. 430 (S.D.N.Y. 1985).............................................................................. 19
Statutes
17 U.S.C. § 102(a) (2006) ................................................................................................. 16
17 U.S.C. § 102(b) (2006) ...................................................................................... 3, 11, 12
17 U.S.C. § 106(2) (2006) ................................................................................................ 17
17 U.S.C. § 107 (2006) ..................................................................................................... 21
Secondary Sources
About, HathiTrust Digital Library,
http://www.hathitrust.org/about (last visited July 3, 2012). ............................................. 10
About the Internet Archive, Internet Archive,
http://archive.org/about/about.php (last visited July 3, 2012) .......................................... 10
Sophia Ananiadou et al., Text Mining and its Potential
Applications in Systems Biology, 24 TRENDS IN BIOTECHNOLOGY 571 (2006) .................. 4

v

Christian Blaschke et al. Information Extraction in Molecular Biology, 3 BRIEFINGS IN
BIOINFORMATICS 154 (2002) .............................................................................................. 4
Patricia Cohen, Digital Keys for Unlocking the Humanities’ Riches, N.Y. TIMES, Nov. 17,
2010, at C1 .......................................................................................................................... 6
Digging Into Data Challenge, http://www.diggingintodata.org/
(last visited July 3, 2012) .................................................................................................. 11
House Report on the 1976 Act, H.R. Rep. No. 94-1476 (1976) ....................................... 18
James M. Hughes, et al., Quantitative Patterns of Stylistic Influence in the Evolution of
Literature, 109 PROC. OF THE NAT’L ACAD. OF SCI. OF THE U.S. 7682 (2012) ................... 8
Matthew Jockers, Macroanalysis: Digital Methods for Literary History (forthcoming
February 2013) .......................................................................................................... 4-5, 7-9
Brian Lavoie & Lorcan Dempsey, Beyond 1923: Characteristics of Potentially In
Copyright Print Books in Library Collections, 15 D-Lib Mag.,
http://www.dlib.org/dlib/november09/lavoie/11lavoie.html ............................................ 24
Pierre N. Leval, Toward A Fair Use Standard, 103 HARV. L. REV. 1105 (1990) ............ 21
MALLET: MAchine Learning for LanguagE Toolkit, http://mallet.cs.umass.edu/ (last
visited July 2, 2012) ............................................................................................................ 7
Mapping the Republic of Letters, https://republicofletters.stanford.edu/ (last visited July 2,
2012) ................................................................................................................................... 6
Jean-Baptiste Michel, Yuan Kui Shen, Aviva Presser Aiden, Adrian Veres, Matthew K.
Gray, The Google Books Team, Joseph P. Pickett, Dale Hoiberg, Dan Clancy, Peter
Norvig, Jon Orwant, Steven Pinker, Martin A. Nowak, and Erez Lieberman Aiden;
Quantitative Analysis of Culture Using Millions of Digitized Books. 331 SCIENCE 176
(2011) .................................................................................................................................. 7
MONK: Metadata Offer New Knowledge, http://www.monkproject.org/
(last visited July 2, 2012) .................................................................................................... 7
Franco Moretti, Graphs, Maps, Trees: Abstract Models for Literary History (2005)........ 5
Toshihide Ono et al., Automated Extraction of Information on Protein–Protein
Interactions from the Biological Literature, 17 BIOINFORMATICS 155 (2001) ................... 4
Jon Orwant, More Researchers Dive into the Digital Humanities, Google Research Blog
(Dec. 20, 2010), http://googleresearch.blogspot.com/2010/12/more-researchers-dive-intodigital.html ........................................................................................................................ 10

vi

Jon Orwant, Our Commitment to Digital Humanities, Google Research Blog (July 14,
2010), http://googleresearch.blogspot.com/2010/07/our-commitment-to-digitalhumanities.html ................................................................................................................. 11
Project Gutenberg, http://www.gutenberg.org/ (last visited July 3, 2012) ....................... 10
Matthew Sag, Copyright and Copy-Reliant Technology, 103 NW. U.L. REV. 1607 (2009) 2
Matthew Sag, Orphan Works as Grist for the Data Mill, 27 BERKELEY TECH. L. J. ___
(forthcoming 2012) ............................................................................................................. 2
Software Environment for the Advancement of Scholarly Research (“SEASR”)
http://seasr.org (last visited June 29, 2012) ........................................................................ 7
Stanford Literary Lab, http://litlab.stanford.edu/ (last visited June 29, 2012) .................... 4
Text Analysis Portal for Research (“TAPoR”), http://www.tapor.ca/portal/portal (last
visited July 2, 2012) ............................................................................................................ 7

vii

STATEMENT OF INTEREST OF AMICI
Amici are professors and scholars who teach, write, and research in the areas of
either digital humanities or the law, and an association that represents Digital Humanities
scholars generally.1 Digital Humanities Amici have an interest in this case because of its
potential impact on their ability to discover and understand, through automated means,
the data in and relationships among textual works. Legal Scholar Amici have an interest
in this case both because of its impact on socially beneficial research, and because of
their interest in the sound development of intellectual property law. Resolution of the
legal issue of copying for non-expressive uses has far-reaching implications for the scope
of copyright protection, a subject germane to Amici’s professional interests and one about
which they have great expertise. Amici speak only to the issue of copying for nonexpressive uses and express no opinion on other aspects of this case. A complete list of
individual amici is attached as Appendix A.
SUMMARY OF ARGUMENT
The significance of this case extends far beyond one discrete product of a single
information technology company. Mass digitization, like that employed by Google, is a
key enabler of socially valuable computational and statistical research (often called “data
mining” or “text mining”). While the practice of data mining has been used for several
decades in traditional scientific disciplines such as astrophysics and in social sciences like
economics, it has only recently become technologically and economically feasible within
the humanities. This has led to a revolution, dubbed “Digital Humanities,” ranging across
subjects like literature and linguistics to history and philosophy. New scholarly endeavors

1 See

Association for Computers and the Humanities, http://www.ach.org/.
1

enabled by Digital Humanities advancements are still in their infancy, but have enormous
potential to contribute to our collective understanding of the cultural, political, and
economic relationships among various collections (or “corpuses”) of works—including
copyrighted works—and with society. The Court’s ruling in this case on the legality of
mass digitization could dramatically affect the future of work in the Digital Humanities.
In ruling on the parties’ motions, the Court should recognize that text mining is a
non-expressive use that presents no legally cognizable conflict with the statutory rights or
interests of the copyright holders. Where, as here, the output of a database—i.e., the data
it produces and displays—is noninfringing, this Court should find that the creation and
operation of the database itself is likewise noninfringing. The copying required to convert
paper library books into a searchable digital database is properly considered a “nonexpressive use” because the works are copied for reasons unrelated to their protectable
expressive qualities; none of the works in question are being read by humans as they
would be if sitting on the shelves of a library or bookstore.
The type of non-expressive use at issue here is common among copy-reliant
technologies: for example, Internet search engines and plagiarism detection software do
not read, understand, or enjoy copyrighted works, nor do they deliver these works
directly to the public. Such platforms copy the works only incidentally, in order to
process them as “grist for the mill”—raw materials that feed various algorithms and
indices. See Matthew Sag, Copyright and Copy-Reliant Technology, 103 NW. U.L. REV.
1607 (2009); Matthew Sag, Orphan Works as Grist for the Data Mill, 27 BERKELEY
TECH. L. J. ___ (forthcoming 2012).

2

Further, generating data about a copyrighted work (often called “metadata”) does
not infringe the original work because, as has been recognized for over a century,
copyright law protects only an author’s original expression, not facts. That a “fact” might
pertain to an expressive works does not change its factual character—or render it an
author’s exclusive intellectual property under the law. Indeed, making such factual
information freely available to all is crucial to the harmony between copyright law and
the First Amendment—hence the existence of rules like the “idea/expression” distinction
(see 17 U.S.C. § 102(b)), the doctrine of scenes à faire, and the “merger” principle.
The act of copying works into a database in order to enable the generation of
metadata about those works should thus be deemed noninfringing. As numerous courts
have found, making intermediate copies that enable socially beneficial noninfringing uses
and/or outputs constitutes a protected “fair use” under Section 107 of the Copyright Act.
See, e.g., A.V. ex rel. Vanderhye v. iParadigms, LLC, 562 F.3d 630, 645 (4th Cir. 2009);
Perfect 10, Inc. v. Amazon.com, Inc., 508 F.3d 1146, 1168 (9th Cir. 2007); Sony
Computer Entm’t, Inc. v. Connectix Corp., 203 F.3d 596, 609 (9th Cir. 2000); Sega
Enters. Ltd. v. Accolade, Inc., 977 F.2d 1510, 1527-28 (9th Cir. 1992). Similarly, the
mass digitization of books for text-mining purposes is a form of incidental or
“intermediate” copying that enables ultimately non-expressive, non-infringing, and
socially beneficial uses without unduly treading on any expressive—i.e., legally
cognizable—uses of the works. The Court should find such copying to be fair use.
ARGUMENT
I.

The Freedom to Make Non-expressive Use of Copyrighted Works is Vital to
the “Progress of Science” in the Digital Humanities

3

Where large-scale electronic text collections are available, advances in
computational power and a proliferation of new text-mining and visualization tools offer
scholars of the humanities the chance to do what biologists, physicists, and economists
have been doing for decades—analyze massive amounts of data.
“Digital Humanities” scholars fervently believe that text mining and the
computational analysis of text are vital to the progress of human knowledge in the current
Information Age. The potential of these non-expressive uses of text has already been
revealed in the life sciences, where researchers routinely use a variety of text-mining
tools to facilitate the search for relevant research across disparate fields and to uncover
previously unnoticed “correlations or associations such as protein-protein interactions
and gene-disease associations.” See Sophia Ananiadou et al., Text Mining and its
Potential Applications in Systems Biology, 24 TRENDS IN BIOTECHNOLOGY 571, 571
(2006) (citing Toshihide Ono et al., Automated Extraction of Information on Protein–
Protein Interactions from the Biological Literature, 17 BIOINFORMATICS 155 (2001) and
Christian Blaschke et al. Information Extraction in Molecular Biology, 3 BRIEFINGS IN
BIOINFORMATICS 154 (2002)).
Similar breakthroughs are on the horizon in the humanities. Traditionally, literary
scholars have relied upon the close and often anecdotal study of select works. Modern
computing power and the mass digitization of texts now permits investigation of the
larger literary record. Just as the power of computing transformed the study of the
genome (before which, years were required to sequence a single gene) into the study of
genomics (in which entire genomes can be sequenced within days), revolutionizing our

4

understanding of biology and medicine, computers are also transforming the humanities
and bringing about vast new intellectual riches there.
Literary analyses of digitized collections, such as those now curated by the
HathiTrust, are at the core of Digital Humanities research. Large-scale quantitative
projects like those being undertaken at the Stanford Literary Lab are unearthing
previously unknowable information about individual works, genres, and even entire eras.2
Digitization enhances our ability to process, mine, and ultimately better understand
individual texts, the connections between texts, and the evolution of literary language. As
University of Nebraska Professor Matthew Jockers explains, by exploring the literary
record writ large, researchers can better understand the context in which individual texts
exist, and thereby better understand the texts themselves. See Matthew Jockers,
Macroanalysis: Digital Methods for Literary History (forthcoming February 2013).
Along similar lines, Stanford University Professor Franco Moretti has noted that “a field
this large cannot be understood by stitching together separate bits of knowledge about
individual cases, because it isn’t a sum of individual cases: it’s a collective system, that
should be grasped as such, as a whole . . . .” Franco Moretti, Graphs, Maps, Trees:
Abstract Models for Literary History 4 (2005).
Researchers working in Information Retrieval frequently use text mining and
computer-aided classification to identify and retrieve relevant documents. Using similar
techniques, researchers in the Digital Humanities are able to identify and retrieve relevant
texts, often from unlikely places. Humanities researchers can thereby expand their

2

The Stanford Literary Lab discusses, designs, and pursues literary research of a digital
and quantitative nature. See Stanford Literary Lab, http://litlab.stanford.edu/ (last visited
June 29, 2012).
5

traditional study of a few canonical works to a study of any one of the several million
books in the larger archive of literary history—an archive that has hitherto remained
hidden because of the limitations of humans’ reading capacity. In the process,
nonexpressive use leads to additional expressive use, expanding the audience (and the
potential market) for individual works.3
Mass digitization also results in the creation of data that enables scholars to
reimagine relationships between texts—for example, by linking texts with maps. Thus,
Google’s “Ancient Places Project” links the text of public domain books like Gibbon’s
Decline and Fall of the Roman Empire to a map of the ancient world.4 The interface
allows the user to browse the books, including the full text, at the same time as she
browses a map. The places mentioned are marked on the map and hyperlinked.5 Similar
maps could be made with reference to works still under copyright—importantly, without
ever making the text of the book available for free viewing. Extracting such data from
texts to create these maps is a quintessential nonexpressive use of the underlying texts

3

For example, Matthew Jockers used text mining and computer aided classification to
identify an overlooked tradition of whaling fiction predating (and arguably informing)
Melville’s writing of Moby Dick. See Jockers, supra.
4

Elton Barker, Eric C. Kansa, Leif Isaksen, GAP: A Neogeo Approach To Classical
Resources, in EUROPEAN CONFERENCE ON COMPLEX SYSTEMS 2010 (Sep. 2010).
5

In a similar vein, researchers at Stanford University have mapped thousands of letters
exchanged during the Enlightenment and thereby devised a theory of how these
individual networks fit into a coherent whole, which the scholars refer to as the “Republic
of Letters.” Mapping the Republic of Letters, https://republicofletters.stanford.edu/ (last
visited July 2, 2012). Such aggregation yields surprising insights: for example, “the
common narrative is that the Enlightenment started in England and spread to the rest of
Europe,” but the relatively low volume of correspondence between London and Paris
suggests otherwise. See Patricia Cohen, Digital Keys for Unlocking the Humanities’
Riches, N.Y. TIMES, Nov. 17, 2010, at C1.
6

that does not implicate any copyright-protected use—let alone infringe the copyrights
of—the works in question.
Google’s “Ngram” tool provides another example of a nonexpressive use enabled
by mass digitization—this time easily visualized. Figure 1, below, is an Ngram-generated
chart that compares the frequency with which authors of texts in the Google Book Search
database refer to the United States as a single entity (“is”) as opposed to a collection of
individual states (“are”). As the chart illustrates, it was only in the latter half of the
Nineteenth Century that the conception of the United States as a single, indivisible entity
was reflected in the way a majority of writers referred to the nation. This is a trend with
obvious political and historical significance, of interest to a wide range of scholars and
even to the public at large. But this type of comparison is meaningful only to the extent
that it uses as raw data a digitized archive of significant size and scope.6
Figure 1: Google Ngram Visualization Comparing Frequency of
“The United States is” to “The United States are”

6

Google Ngram is available at http://books.google.com/ngrams. This particular ngram
can be reproduced as follows:
http://books.google.com/ngrams/graph?content=The+United+States+is%2C+The+United
+States+are&year_start=1780&year_end=1900&corpus=5&smoothing=10.
7

To be absolutely clear, 1) the data used to produce this visualization can only be
collected by digitizing the entire contents of the relevant books, and 2) not a single
sentence of the underlying books has been reproduced in the finished product. In other
words, this type of nonexpressive use only adds to our collective knowledge and
understanding, without in any way replacing, damaging the value of, or interfering with
the market for, the original works.7
Google Ngram is just the tip of the iceberg.8 In Macroanalysis: Digital Methods
and Literary History, Professor Jockers draws on a corpus of Nineteenth Century novels
to demonstrate how literary style changes over time. See generally Jockers, supra.
Examining word frequencies, syntactic patterns, and thematic markers in the metadataenriched context of author nationality, author gender, and time period, opens up literary
study to an entirely new perspective.9 Trendsetters and outliers are revealed, as when

7

For additional examples of Ngram’s uses, see, e.g., Jean-Baptiste Michel, Yuan Kui
Shen, Aviva Presser Aiden, Adrian Veres, Matthew K. Gray, The Google Books Team,
Joseph P. Pickett, Dale Hoiberg, Dan Clancy, Peter Norvig, Jon Orwant, Steven Pinker,
Martin A. Nowak, and Erez Lieberman Aiden; Quantitative Analysis of Culture Using
Millions of Digitized Books. 331 SCIENCE 176 (2011) (a study of linguistic and cultural
changes in over five million digitized books).
8 The

toolkit available to Digital Humanities researchers is becoming increasingly
sophisticated. See, e.g., Text Analysis Portal for Research (“TAPoR”),
http://www.tapor.ca/portal/portal (last visited July 2, 2012) (tools to map word usage over
time, including peaks, density, collocations, and types); MALLET: MAchine Learning
for LanguagE Toolkit, http://mallet.cs.umass.edu/ (last visited July 2, 2012) (a Java-based
package for statistical natural language processing, document classification, clustering,
topic modeling, information extraction, and other machine learning applications to text);
MONK: Metadata Offer New Knowledge, http://www.monkproject.org/ (last visited July
2, 2012) (a digital environment designed to help humanities scholars discover and
analyze patterns in the texts); Software Environment for the Advancement of Scholarly
Research (“SEASR”), http://seasr.org (last visited June 29, 2012).
9

A recently published study, led by mathematicians at Dartmouth, makes a similar point.
See James M. Hughes et al., Quantitative Patterns of Stylistic Influence in the Evolution
of Literature, 109 PROC. OF THE NAT’L ACAD. OF SCI. OF THE U.S. 7682 (2012).
8

Jockers’ text mining and computational analysis demonstrated that Harriet Beecher
Stowe’s fiction is far more similar to the work of male authors of her generation than to
the female-authored works of “sentimental fiction” among which her work has
traditionally been categorized. See Jockers, supra.
Figure 2 provides another fascinating example of Professor Jockers’ research.
The chart shows the extent to which British, American, and Irish authors focused on the
theme of American slavery during the Nineteenth Century, based on a corpus of 3,450
novels from that time period. Although it comes as no surprise that slavery was most
often addressed by American authors, the strong Irish reaction to the American Civil War
(note the spike in the light gray line beginning in 1860) compared with the decidedly
muted response by British authors invites—indeed, demands—further investigation.
Figure 2: American Slavery in American, English, and Irish Literature, 1800-1899.

9

As Jockers’ work reveals, “macroanalysis” of text archives has the potential to
provide insight into historical literary questions, such as the place of individual texts,
authors, and genres in relation to a larger literary context; literary patterns and lexicons
employed over time, across periods, within regions, or within demographic groups; the
cultural and societal forces that impact literary style and the evolution of style; the
waxing and waning of literary themes; and the tastes and preferences of the literary
establishment—and whether those preferences correspond to general tastes and
preferences. However, realizing this potential requires access to digitized texts.
If libraries, research universities, non-profit organizations,10 and commercial
entities like Google11 are prohibited from making nonexpressive use of copyrighted
material, literary scholars, historians, and other humanists are destined to become 19thcenturyists; slaves not to history, but to the public domain. History does not end in
1923.12 But if copyright law prevents Digital Humanities scholars from using more recent
materials, that is the effective end date of the work these scholars can do.
In short, the possibility of mining huge digital archives and manipulating the data
collected in the process has inspired many scholars to reconceptualize the very nature of

10

See, e.g., HathiTrust Digital Library, http://www.hathitrust.org/about (last visited July
3, 2012); Project Gutenberg, http://www.gutenberg.org/ (last visited July 3, 2012); About
the Internet Archive, http://archive.org/about/about.php (last visited July 3, 2012).
11 Google

has played a significant role in facilitating research in the Digital Humanities,
but the legality of nonexpressive use is an issue that transcends any one digital repository.
In 2009, an international coalition of research organizations including the National
Science Foundation, the Institute of Museum and Library Services, and the National
Endowment for the Humanities sponsored a multi-million dollar competition to promote
innovative humanities and social science research using large-scale data analysis. See
Digging Into Data Challenge, http://www.diggingintodata.org/ (last visited July 3, 2012).
12 Due

to repeated extensions of the copyright term, U.S. copyrights after 1923 do not
automatically expire on an annual basis; thus, most modern works are still copyrighted.
10

humanities research. For others, it has played the more modest—but still valuable—role
of providing new tools for testing old theories, or suggesting new areas of inquiry. None
of this, however, can be done in the Twentieth-Century context if scholars cannot make
nonexpressive uses of underlying copyrighted texts, which (as shown above) will
frequently number in the thousands, if not millions. Given copyright law’s objective of
promoting “the Progress of Science,”13 it would be perversely counterintuitive if the
promise of Digital Humanities were extinguished in the name of copyright protection.
II.

COPYRIGHT LAW DOES NOT PROTECT NONEXPRESSIVE ASPECTS
OF WORKS
Fortunately, this Court need not contemplate such a scenario, as nonexpressive

aspects of copyrighted works—e.g., the facts and ideas contained within the work and
concerning it—are not protected by copyright. Such fundamental legal principles as the
“idea/expression” distinction (reflected in Section 102(b) of the Copyright Act), the
“merger” doctrine, the rule of “scènes à faire,” and the “fact/expression” distinction all
reflect this basic tenet. Metadata—information about copyrighted works collected
through data mining and used by Digital Humanities scholars in the research described
above—either does not implicate copyright protection at all, or is inoculated by the
aforementioned doctrines that limit authors’ rights to their works’ expressive content.
A.

The Idea/Expression Distinction

Copyright gives authors the right to set the terms upon which their original
expression is made available to the public. But this right is not unlimited. As one of the
fundamental—and Constitutional—limitations on those rights, the idea-expression

13

U.S. Const. Art I., Sec. 8.
knowledge and learning.

“Science,” as used in the Constitution, referred to
11

distinction strikes a balance between “the interests of authors . . . in the control and
exploitation of their writings . . . on the one hand, and society’s competing interest in the
free flow of ideas, information, and commerce on the other hand.” Harper & Row
Publishers, Inc. v. Nation Enters., 471 U.S. 539 (1985) (quoting Sony Corp. of Am. v.
Universal City Studios, Inc., 464 U.S. 417, 429 (1984)); see also Golan v. Holder, 132 S.
Ct. 873, 890 (2012) (describing the idea-expression distinction as one of copyright’s
“built-in First Amendment accommodations”). Copyright law protects only expressive
use: “It is an axiom of copyright law that the protection granted to a copyrightable work
extends only to the particular expression of an idea and never to the idea itself.” Reyher v.
Children’s Television Workshop, 533 F.2d 87, 90 (2d Cir. 1976).
B.

Section 102(b)

Recognizing the importance of access to ideas within expressive works, Congress
has placed statutory limits on the rights of copyright holders through Section 102(b) of
the Copyright Act, which provides: “In no case does copyright protection for an original
work of authorship extend to any idea . . . concept, principle, or discovery, regardless of
the form in which it is described, explained, illustrated, or embodied in such work.” 17
U.S.C. § 102(b) (2006). This provision has played a key role in modern copyright cases,
ensuring that access to nonexpressive aspects of works is not inhibited. See, e.g., Peter F.
Gaito Architecture, LLC v. Simone Dev. Corp., 602 F.3d 57, 67 (2d Cir. 2010) (holding
that the principle behind § 102(b) required the court “to determine whether . . .
‘similarities are due to protected aesthetic expressions original to the allegedly infringed
work, or whether the similarity is to something in the original that is free for the taking’”
(quoting Tufenkian Imp./Exp. Ventures, Inc. v. Einstein Moomjy, Inc., 338 F.3d 127, 134-

12

35 (2d Cir. 2003))). As noted above, text mining extracts and compiles ideas, concepts,
and principles in copyrighted works into metadata. This process generates the very types
of “discovery” that § 102(b) envisions.
C.

Merger and Scènes à Faire

The policy of excluding nonexpressive elements from copyright protection is so
strong that—even in situations where expressive and nonexpressive elements
intertwine—doctrines like that of “merger” and “scènes à faire” preclude copyright
protection for expression “in those instances where there is only one or so few ways of
expressing an idea that protection of the expression would effectively accord protection
to the idea itself.” Kregos v. Associated Press, 937 F.2d 700, 705 (2d Cir. 1991); New
York Mercantile Exch., Inc. v. IntercontinentalExchange, Inc., 497 F.3d 109, 118 (2d Cir.
2007). The “merger” doctrine is built upon the same principle as the idea/expression
distinction: the protection of expressive elements of a work cannot, for Constitutional and
practical reasons, interfere with the public’s “free access to ideas.” New York Mercantile
Exch., Inc., 497 F.3d. at 116. Relatedly, elements of a work that are scènes à faire—that
is, “incidents, characters or settings which are as a practical matter indispensable, or at
least standard, in the treatment of a given topic”—are not protectable. Hoehling v.
Universal City Studios, Inc., 618 F.2d 972, 979 (2d Cir. 1980); MyWebGrocer, LLC v.
Hometown Info, Inc., 375 F.3d 190, 194 (2d Cir. 2004) (“Scènes à faire are unprotectible
elements [following] from a work’s theme rather than from an author’s creativity.”).
D.

Fact/Expression Distinction

Finally, the monopoly rights of authors cannot extend to factual elements that “do
not owe their origin to an act of authorship.” Feist Publ’ns, Inc. v. Rural Tel. Serv. Co.,

13

Inc., 499 U.S. 340, 347 (1991). “The distinction is one between creation and discovery:
The first person to find and report a particular fact has not created the fact; he or she has
merely discovered its existence.” Id. The Supreme Court in Feist made clear that if an
“author clothes facts with an original collocation of words, he or she may be able to claim
a copyright in this written expression;” nevertheless, “[o]thers may copy the underlying
facts from the publication . . . .” Id. at 348.
In National Basketball Association v. Motorola, Inc., 105 F.3d 841 (2nd Cir.
1997), for example, a sports reporting service distributing real-time game statistics based
on a data feed from reporters was held non-infringing. The Second Circuit reasoned that
“[b]ecause [the service reproduced] only factual information culled from the broadcasts
and none of the copyrightable expression of the games, appellants did not infringe the
copyright of the broadcasts.” Id. at 847. The Second Circuit has similarly held that one
has “the right to avail himself of the facts contained in [another’s] book and to use such
information, whether correct or incorrect, in his own literary work.” Hoehling, 618 F.2d
at 979 (quotations omitted). In other words, copyright law clearly distinguishes between
expressive and nonexpressive content, and deems only expressive content protectable.
E.

Nonexpressive Metadata Does Not Implicate the Statutory Rights of
the Copyright Holder

Metadata about a copyrighted work does not implicate any legally cognizable
interest of the copyright holder. Metadata may contain facts about the works themselves,
might capture (in different terminology) the ideas contained within the text, or may
convey information such as the number of times a given word appears in a particular text,
how often a particular author uses a specific literary device, or the essence of what the

14

work is about.14 Though it is true that metadata would not exist but for the underlying
work, it does not contain the expression of the work.
Consider, for example, two facts about Moby Dick: first, that the word “whale”
appears 1119 times; second, that the word “dinosaur” appears 0 times. While a whale is
certainly central to the expression contained in Moby Dick, this data is not. Rather,
metadata of this sort—a simplified version of the metadata surveyed in Section I—is
factual and nonexpressive, and incapable of infringing the rights of copyright holders.
The same principle can be illustrated using a recent decision of this court, Warner
Brothers Entertainment Inc. v. RDR Books, 575 F. Supp. 2d 513 (S.D.N.Y. 2008).
Consider the following four statements:
[1] “Goblin-made armour does not require cleaning, simple girl. Goblins’ silver
repels mundane dirt, imbibing only that which strengthens it.”
[2] “goblin-made armor does not require cleaning, because goblins’ silver repels
mundane dirt, imbibing only that which strengthens it, such as basilisk venom.”
[3] “Statement [1] contains twenty words, and other than ‘Goblin’, no word in
expression [1] is repeated.”
[4] “Statement [2] is strikingly similar to Statement [1].”
Statement [1] originates with J.K. Rowling, the author of the Harry Potter novels.
See Warner Bros., 575 F. Supp. 2d at 527 (quoting J.K. Rowling, Harry Potter and the
Deathly Hallows 303 (2007)). Statement [2] was held out as originating with a
contributor to the Harry Potter Lexicon (a reference work for the “Harry Potter
universe”), which was found to infringe because too much of its contents consisted of

14

See supra, Part I for examples of the use of such metadata.
15

direct quotations or close paraphrases of vivid passages in the Harry Potter books, as the
comparison between [1] and [2] illustrates. Id. at 527. Statements [3] and [4], by contrast,
are classic metadata; they would not exist but for the underlying work, and yet neither
passage is substantially similar—or indeed, bears any resemblance at all—to the
expressive elements of the underlying work.
Even more importantly, this metadata does not originate with the author of the
underlying work. As the Supreme Court held in Feist Publications, “copying of
constituent elements of the work that are original” is an essential element of a copyright
infringement claim. 499 U.S. at 361 (emphasis added); see also 17 U.S.C. § 102(a)
(2006) (copyright subsists only in “original works of authorship”) (emphasis added).
Amici wish to emphasize that metadata is not the same thing as so-called
“invented facts.” J.K. Rowling’s conception and description of goblin armor and
thousands of other details in the Harry Potter series could be regarded as “invented facts”
because, quite simply, she made them up. As laid out in the case law, if such facts and
their associated expressive descriptions are reproduced in sufficient quantity, they may
“constitute creative expression protected by copyright because characters and events
spring from the imagination of the original authors.” Warner Bros., 575 F. Supp. 2d at
536 (quoting Castle Rock Entm’t Inc. v. Carol Publ’g Grp., Inc., 150 F.3d 132, 139 (2d
Cir. 1998)). Metadata, however, cannot be accurately characterized as “invented facts,”
but only as facts about “invented facts.” The distinction is significant: once again, facts
are not eligible for copyright protection.
Nor does metadata of the sort described above not infringe the author’s right “to
prepare derivative works based upon the copyrighted work[.]” 17 U.S.C. § 106(2) (2006).

16

As this Court held in Warner Brothers, an analytical work that provides insight into an
copyrighted work but does not “recast, transform, or adapt” that work does not violate the
derivative work right. 575 F. Supp. 2d at 539; see also Ty, Inc. v. Publ'ns Int’l Ltd., 292
F.3d 512, 520 (7th Cir. 2002) (holding that collectors’ guide to certain copyrighted works
did not violate 17 U.S.C. § 106(2) because the guides did not “recast, transform, or adapt
the things to which they are guides”).
Amici urge the Court to carefully distinguish the facts of the instant case from
those in Castle Rock Entertainment v. Carol Publishing Group, 150 F.3d 132 (2d Cir.
1998). In Castle Rock, the Second Circuit held that a quiz book based on the popular
television series “Seinfeld” was, quantitatively and qualitatively, substantially similar to
that series, considered as a whole. Id. at 138–39. The quiz book in that case, however,
was not an analytical work; rather, it essentially recast “Seinfeld”’s copyrightable
characters into a new format, as if the defendant had made miniature dolls of those same
characters. See Hasbro Bradley, Inc. v. Sparkle Toys, Inc., 780 F.2d 189, 192-93 (2d Cir.
1985) (upholding copyrightability of “Transformer” robotic action figures as sculptural
works). The supposed “facts” conveyed in the “Seinfeld” quiz book were not truly facts
about the television program; they were “in reality fictitious expression created by
Seinfeld’s authors.” Castle Rock Entm’t, 150 F.3d at 139.
By contrast, the many forms of metadata produced by the library digitization at
the heart of this litigation do not merely recast copyrightable expression from underlying
works; rather, the metadata encompasses numerous uncopyrightable facts about the
works, such as author, title, frequency of particular words or phrases, and the like.
F.

Nonexpressive Metadata Is Also Noninfringing Because It Does Not
Allow the Public to Perceive the Expressive Content of a Work
17

The significance of public perception runs deep in copyright law. Indeed,
controlling authority suggests that the copyright holder’s exclusive rights are limited to
the right to communicate the expressive aspects of her work to the public. For example,
in New York Times Co. v. Tasini, 533 U.S. 483 (2001), a case about the scope of the 17
U.S.C. § 201(c) “privilege” of the copyright owner to reproduce and distribute individual
contributions “as part of [a] collective work,” the Supreme Court held that “[i]n
determining whether the Articles [at issue] have been reproduced and distributed as part
of a revision of the collective works in issue, we focus on the Articles as presented to,
and perceptible by, the user[s] of the Databases [containing the Articles].” 533 U.S. at
499 (emphasis added; internal quotation marks and citations omitted). The Court
elaborated: “the question is not whether a user can generate a revision of a collective
work from a database, but whether the database itself perceptibly presents the author’s
contribution as part of a revision of the collective work.” Id. at 504 (emphasis added).15
This point is especially evident in cases where plaintiffs have argued that,
although a defendant’s final product does not support an allegation of infringement, the
defendant has violated the Copyright Act by making a reproduction of the plaintiff’s
work that is merely intermediate, transient, and imperceptible by the public. In Davis v.
United Artists, Inc., for example, this Court rejected out of hand the allegation that the
defendant’s unpublished screenplays were substantially similar to plaintiff’s novel,

15

It is true that the text of Section 106(1) of the Copyright Act, the reproduction right,
has no express limitation of this sort, yet courts have focused their analysis in this way.
Even where courts conclude that reproduction without public reception constitutes a
prima facie case of infringement, the question of public reception remains important. See
Cambridge Univ. Press v. Becker, No. 08 Civ. 01425, Slip Op. at 99–108, 133, 203, 235,
283 (N.D. Ga. May 11, 2012) (holding that reproduction and posting of works online,
without subsequent viewing by users, was de minimis use and therefore not infringing).
18

refusing to “consider the preliminary scripts” because “the ultimate test of infringement
must be the film as produced and broadcast” to the public. 547 F. Supp. 722, 724 n.9
(S.D.N.Y. 1982). See also Fuld v. Nat’l Broad. Co., Inc., 390 F. Supp. 877, 882 n.4
(S.D.N.Y. 1975) (“[T]he ultimate test of infringement must be the television film as
produced and broadcast — and not the preliminary scripts . . . .”); Walker v. Time Life
Films, Inc., 615 F. Supp. 430, 434 (S.D.N.Y. 1985) (“The Court considers the works as
they were presented to the public.”)16
III.

Text Mining Creates Value by Facilitating the Advancement of Our
Collective Knowledge; To Protect That Value, Mass Digitization and Similar
Intermediate Copying For Data Mining and Other Nonexpressive Purposes
Should Be Considered "Fair Use"
As demonstrated above, nonexpressive metadata itself is noninfringing. However,

Amici recognize that this Court must also consider the legality of the process of making
copies to generate that metadata. Fortunately, numerous courts have held that copying to
enable purely nonexpressive uses, such as the automated extraction of data, does not
infringe the statutory rights of the copyright holder. Like copying employed for other
“transformative” purposes, such as parody, criticism, and reverse engineering,
intermediate copying for the purpose of extracting nonexpressive metadata is fair use.
A.

Nonexpressive Copying to Expand Our Knowledge in the Digital
Humanities Is An Activity of the Sort that Copyright Law Should
Favor, Through Fair Use

16 Courts

in other circuits have adopted the same view. See, e.g., Stromback v. New Line
Cinema, 384 F.3d 283, 299 (6th Cir. 2004) (“In deciding infringement claims, courts
have held that only the version of the alleged infringing work presented to the public
should be considered”); Madrid v. Chronicle Books, 209 F. Supp. 2d 1227, 1234 (D. Wyo.
2002) (“Since a court considers the works as they were presented to the public, discovery
in this case . . . would be pointless”) (internal quotation marks omitted).
19

First among the statutory factors relevant to a fair use analysis is “the purpose and
character of the use, including whether such use is of a commercial nature or is for
nonprofit educational purposes.” 17 U.S.C. § 107(1). The more “nonexpressive” the use
of a copyrighted work, the less it substitutes for the author’s original expression. As such,
nonexpressive uses are properly considered equivalent (though not identical) to highly
transformative uses: their “purpose and character” is such that they do not merely
supersede the objects of the original creation. Campbell v. Acuff-Rose Music, Inc., 510
U.S. 569, 583 (1994). See also Perfect 10, Inc. v. Amazon.com, Inc., 508 F.3d 1146,
1165 (9th Cir. 2007) (holding that search engines are “highly transformative” because
“[a]lthough an image may have been created originally to serve an entertainment,
aesthetic, or informative function, a search engine transforms the image into a pointer
directing a user to a source of information”); Kelly v. Arriba Soft Corp., 336 F.3d 811,
818 (9th Cir. 2002) (holding that use of images in search engine was transformative
because they served “as a tool to help index and improve access to images on the internet
and their related web sites” and their use was “unrelated to any aesthetic purpose”); Bill
Graham Archives v. Dorling Kindersley Ltd., 448 F.3d 605, 609 (2d Cir. 2006) (finding
critical to fair use analysis that publisher’s use of copyrighted images of concert posters
in book was “plainly different from the original purpose for which they were created”).
As the process of digitization for text mining is intermediate and nonexpressive, and its
purpose is to produce nonexpressive metadata, this factor favors fair use.
Moreover, “there is a strong presumption that factor one [in the fair use analysis]
favors the defendant if the allegedly infringing work fits the description of uses described
in [17 U.S.C.] § 107,” which includes “scholarship” and “research.” NXIVM Corp. v.

20

Ross Institute, 364 F.3d 471, 477 (2d Cir. 2004). The crucial role that mass digitization
plays in promoting the progress of research and scholarship in the Digital Humanities
weighs heavily in favor of fair use here. See also Pierre N. Leval, Toward A Fair Use
Standard, 103 HARV. L. REV. 1105, 1111 (1990) (“If [a] secondary use adds value to the
original – if the quoted matter is used as raw material, transformed in the creation of new
information, new aesthetics, new insights and understandings – this is the very type of
activity that the fair use doctrine intends to protect for the enrichment of society.”)
Similarly, courts have ruled in favor of fair use when copying allowed defendants
or third parties to use facts from copyrighted works in news reporting or court
proceedings. See, e.g., Bond v. Blum, 317 F.3d 385, 395 (4th Cir. 2003) (holding that “the
narrow purpose of defendants’ use of the manuscript . . . for the evidentiary value of its
content” weighed “heavily” against a finding of infringement); Religious Tech. Ctr. v.
Lerma, 908 F. Supp. 1362, 1366 (E.D. Va. 1995) (finding fair use in part because
documents were copied for “news gathering, news reporting and responding to litigation,”
not to “scoop” copyright owner). Significantly, both the Bond and Religious Tech. Ctr.
courts’ fair use holdings went further than the text mining at issue here, because the users
in those cases had to glean the necessary facts by reading the materials, rather than
mining the text with computers. Bond, 317 F.3d at 393; Religious Tech. Ctr., 908 F. Supp.
at 1364-65. If a human’s reading of copyrighted expression to extract nonexpressive
material is fair use, the result must be the same when a computer performs the extraction.
Finally, the commercial nature of Google Inc. should not weigh against a finding
of fair use where, as here, its scanning of academic library holdings enables
noninfringing uses that promote one of copyright’s central purposes: the expansion of

21

knowledge. See, e.g., A.V. ex rel. Vanderhye v. iParadigms, LLC, 562 F.3d 630, 638 (4th
Cir. 2009) (finding that commercial character does not weigh heavily against fair use
when there is a transformative or publicly beneficial purpose); Perfect 10, Inc. v.
Amazon.com, Inc., 508 F.3d 1146, 1166 (9th Cir. 2007) (same); Sony Computer Entm't,
Inc. v. Connectix Corp., 203 F.3d 596, 607 (9th Cir. 2000) (same); Sega Enters. Ltd. v.
Accolade, Inc., 977 F.2d 1510, 1522-23 (9th Cir. 1992) (same). As the Supreme Court
has recognized, “[if] commerciality carried presumptive force against a finding of
fairness, the presumption would swallow nearly all of the illustrative uses listed in the
text of § 107, including news reporting, comment, criticism, teaching, scholarship, and
research, since these activities ‘are generally conducted for profit in this country.’”
Campbell, 510 U.S. at 584 (internal citations omitted). Use of the Google scans for
scholarship and research is very the type of commercial fair use described in Campbell.
B.

The Nature of the Works in Question Is Neutral to the Fair Use
Analysis of Mass Digitization for the Advancement of Digital
Humanities Research and Scholarship

When the purpose of a secondary use is socially beneficial, the second fair use
factor, “the nature of the copyrighted work,” is rarely dispositive. See, e.g., Bill Graham,
448 F.3d at 612 (“The second factor may be of limited usefulness where the creative
work of art is being used for a transformative purpose.”) This is especially true in
“intermediate copying” cases like this one, where the material ultimately reaching the
user is not the expressive content of the copyrighted work at all, but rather ideas
contained within it or facts about it.
Nevertheless, to the extent that the second fair use factor is relevant here, it
weighs in favor of fair use. Looking to this factor, “[c]ourts generally hold that ‘the scope

22

of the second fair use is greater with respect to factual than non-factual works’. . . .
[F]ictional works, on the other hand, . . . require more protection.” Basic Books, Inc. v.
Kinko's Graphics Corp., 758 F. Supp. 1522, 1533 (S.D.N.Y. 1991) (quoting New Era
Publications Int'l, ApS v. Carol Pub. Group, 904 F.2d 152, 157 (2d Cir. 1990)). A
detailed study of the copyrighted works in the collections from which Google has created
its digitized corpus have concluded that the “overwhelming majority – 92 Percent . . . –
were non fiction.” Brian Lavoie & Lorcan Dempsey, Beyond 1923: Characteristics of
Potentially In Copyright Print Books in Library Collections, 15 D-Lib Mag.,
http://www.dlib.org/dlib/november09/lavoie/11lavoie.html.
Furthermore, as one court explained, the second fair use factor weighs in favor of
fair use where humans “cannot gain access to the unprotected ideas and functional
concepts contained in [the copyrighted work] without . . . making copies.” Sega, 977 F.2d
at 1525. This is effectively the case for Digital Humanities scholars, as there are no
plausible ways to conduct analyses of the sort described in Section I other than mass
digitization and algorithmic analysis, both of which require making intermediate copies.
C.

To the Extent Relevant, Mass Digitization Uses a Reasonable
“Amount and Substantiality” of the Works in Question, in Light of
the Socially Beneficial Purpose of Facilitating Data Mining for the
Advancement of the Digital Humanities

The third fair use factor asks whether the amount and substantiality used are
“reasonable in relation to the purpose of the copying.” Campbell, 510 U.S. at 586–87.
Because the metadata created here does not contain any infringing material, the third
factor “is of very little weight.” See, e.g., Connectix, 203 F.3d at 606. This is true even
where many intermediate copies are made. Id. at 601. Moreover, as Section I shows, it is
not only reasonable to use mass digitization of an entire set of works to enable the
23

creation of noninfringing metadata about those works, it is a practical necessity, as there
is no equivalent human means of doing so. In order for Digital Humanities research and
scholarship to be as accurate and complete as possible, every word or image in a
copyrighted work must be mined.
Other courts have relied upon similar rationales to support full copying in
intermediate and nonexpressive fair use cases. See, e.g., Vanderhye, 562 F.3d at 642
(finding mass digitization of entire student essays to be fair use when reasonable as a
means to check for plagiarism); Perfect 10, 508 F.3d at 1167-68 (finding thumbnail
reproduction of entire photographs reasonable in light of defendant’s use of the images to
improve access to information on the internet versus artistic expression); Kelly, 336 F.3d
820-21 (same); Bond, 317 F.3d at 396 (noting that “[t]he use of the copyrighted material
[as evidence in a custody proceeding], even the entire manuscript, does not undermine
the protections granted by the [Copyright] Act”). In light of practical necessity and ample
precedent in support, the Court should find that the “amount and substantiality” factor
favors the making of intermediate copies for the purpose of scholarship and research.
D.

Allowing Intermediate Copying in Order to Enable Nonexpressive
Uses Does Not Harm the Market for the Original Works in a Legally
Cognizable Manner, As The Practice Does Not Implicate the Works'
Expressive Aspects in Any Way

The fourth statutory fair use factor is “the effect of the use upon the potential
market for or value of the copyrighted work.” In the case of expressive uses such as
parody, and nonexpressive uses such as reverse engineering, courts have consistently
held that the protection that copyright affords is limited to certain cognizable markets.
Campbell, 510 U.S. at 591-92 (“[W]hen a lethal parody, like a scathing theater review,
kills demand for the original, it does not produce a harm cognizable under the Copyright
24

Act.”); Sega, 977 F.2d at 1523-24. Transformative expressive uses do not usually affect
the market in any relevant sense because the second author’s expression does not
substitute for that of the original author. Campbell, 510 U.S. at 591; Fisher v. Dees, 794
F.2d 432, 438 (9th Cir. 1986) (“This is not a case in which commercial substitution is
likely . . . . The two works do not fulfill the same demand.”). As illustrated by the
examples in Section I, above, nonexpressive uses have no potential substitution effect on
any legally cognizable market for copyrighted works, because copyright only protects
markets for expression, and not markets for discoveries, ideas, facts, principles, or
concepts. See, e.g., Vanderhye, 562 F.3d at 644 (“[N]o market substitute was created by
[defendants], whose archived student works do not supplant the plaintiffs’ works . . . so
much as merely suppress demand for them . . . In our view, then, any harm here is not of
the kind protected against by copyright law.”) Indeed, in many instances, the use of
metadata made by scholars could actually enhance the market for the underlying work, by
causing researchers to revisit the original work and reexamine it in more detail.
In short, there is simply no reason not to allow the practice at issue, which holds
great promise for Amici, other scholars, society at large—and copyright owners, too.
DATED: July 6, 2012
New York, New York

Respectfully Submitted,

_______________________________
CHARLES COLMAN LAW, PLLC
Charles E. Colman (CC-1133)
30 West 70th Street
New York, NY 10023
Phone: [phone]
Facsimile: 212-960-8969
E-mail: [e-mail]
Counsel for Amici (with Profs. Sag and Schultz)
25

APPENDIX A
Erez Lieberman Aiden
Fellow
Harvard Society of Fellows
Harvard University
Steve Anderson
Professor
School of Cinematic Arts
University of Southern California
The Association for Computers and the Humanities
http://www.ach.org
Dr. Elton Barker
Department of Classical Studies
The Open University, UK
Ann Bartow
Professor of Law
Pace Law School
Jeremy Boggs
Scholars’ Lab
University of Virginia Library
Annemarie Bridy
Fellow and Visiting Associate Research Scholar
Center for Information Technology Policy (CITP)
Princeton University
Susan Brown
Professor
School of English and Theatre Studies
University of Guelph
Director, Orlando Project; Project Leader, Canadian Writing Research Collaboratory
Dan L. Burk
Chancellor’s Professor of Law
UC Irvine School of Law
Dr. Kate Byrne
School of Informatics
University of Edinburgh

26

Prof. Dr. Irene Calboli
Director, Intellectual Property and Technology Program
Marquette University Law School
Alexandra Chassanoff
Doctoral Student
School of Information and Library Science, University of North Carolina at Chapel Hill
Penelope Eckert
Professor of Linguistics
Professor by Courtesy of Anthropology
Stanford University
Laura N. Gassaway
Paul B. Eaton Distinguished Professor of Law
University of North Carolina School of Law
David Theo Goldberg
Director and Professor
University of California
Humanities Research Institute
Eric Goldman
Director, High Tech Law Institute
Santa Clara University School of Law
Dr. Leif Isaksen
Lecturer in Digital Humanities
University of Southampton
Matthew Jockers
Assistant Professor of English
Fellow, Center for Digital Humanities Research
University of Nebraska, Lincoln
Faye E. Jones
Professor
The Florida State University College of Law
Dan Jurafsky
Professor of Linguistics
Professor by Courtesy of Computer Science
Stanford University
Eric Kansa
UC Berkeley School of Information
27

Alexandria Archive Institute
Dennis S. Karjala
Jack E. Brown Professor of Law
Sandra Day O'Connor College of Law
Arizona State University
Professor Jarom McDonald
Director, Office of Digital Humanities
Brigham Young University
Jean-Baptiste Michel
Postdoctoral Fellow
Harvard University
Franco Moretti
Professor of English
Stanford University
Lateef Mtima
Professor of Law
Howard University
Martin Mueller
Professor Emeritus of English, Classics, and Comparative Literature
Northwestern University
Dr. Bethany Nowviskie
University of Virginia Library
Frank A. Pasquale
Schering-Plough Professor in Health Care Regulation and Enforcement
Seton Hall Law School
Aaron Perzanowski
Assistant Professor
Wayne State University Law School
Malla Pollack
Co-Author, Callmann on Unfair Competition, Trademarks & Monopolies (4th ed.
Thomson-Reuter)
Alex H. Poole
Doctoral Student
University of North Carolina at Chapel Hill

28

Kenneth M. Price
Co-director, Walt Whitman Archive
Hillegass University Professor
Department of English
University of Nebraska-Lincoln
Ivan A. Sag
Sadie Dernham Patek Professor in Humanities and Professor of Linguistics
Stanford University
Pamela Samuelson
Richard M. Sherman Distinguished Professor of Law and Professor of Information
UC Berkeley School of Law and School of Information
Ted Sichelman
Professor
University of San Diego School of Law
Jessica Silbey
Professor of Law
Suffolk University Law School
Rebecca Tushnet
Professor of Law
Georgetown Law
Ted Underwood
Associate Professor of English
University of Illinois, Urbana-Champaign
Thomas Wasow
Professor of Linguistics
Clarence Irving Lewis Professor of Philosophy
Stanford University
Matthew Wilkens
Professor of English
University of Notre Dame
Glen Worthey
Digital Humanities Librarian
Stanford University
Vika Zafrin
Institutional Repository Librarian
Boston University
29

30