Monthly Archives: November 2013

United States Is versus United States Are

When Matthew Jockers, Jason Shultz and I were writing the Digital Humanities Amicus Briefs relating to the Google Books and HathiTrust cases, we searched for an illustration that would concisely explain why data mining expressive works was (a)  socially valuable and (b)  no threat to the copyright interests of the authors of the underlying works. We came across a graph produced using the Google n-gram tool that perfectly fit the bill. The graph below was part of the Digital Humanities Amicus Brief in both the HathiTrust and Google Books cases.



This graph is a reconstruction of data generated using Google Ngram, sampled at five-year intervals. The y-axis is scaled to 1/100,000 of a percent, such that 1 = 0.00001%.

The graph was referred to by the District Court in Authors Guild v. HathiTrust and last week’s decision in Authors Guild v. Google. As we explained in our brief, “[the figure] 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.”

Metadata like this can only be collected by digitizing the entire contents of books, and it clearly does not communicate any author’s original expression to the reading public.

I decided that the graph deserved its own post.

Google Books held to be fair use

Authors Guild v. Google: library digitization as fair use vindicated, again.

After more than eight years of litigation, the legality of the Google Books Search engine has finally been vindicated.

Screen Shot 2013-11-14 at 10.35.00 AM

Authors Guild v Google Summary Judgement (Nov. 14, 2013)

The heart of the decision

The key to understanding Authors Guild v. Google is not in the court’s explanation of any of the individual fair use factors — although there is a great deal here for copyright lawyers to mull over —  but rather in the court’s description of its overall assessment of how the statutory factors should be weighed together in light of the purposes of copyright law.

“In my view, Google Books provides significant public benefits. It advances the progress of the arts and sciences, while maintaining respectful consideration for the rights of authors and other creative individuals, and without adversely impacting the rights of copyright holders. It has become an invaluable research tool that permits students, teachers, librarians, and others to more efficiently identify and locate books. It has given scholars the ability, for the first time, to conduct full-text searches of tens of millions of books. It preserves books, in particular out-of-print and old books that have been forgotten in the bowels of libraries, and it gives them new life. It facilitates access to books for print-disabled and remote or underserved populations. It generates new audiences and creates new sources of income for authors and publishers. Indeed, all society benefits.”  (Authors Guild v. Google, p.26)

Even before last year’s HathiTrust decision (Authors Guild v. Hathitrust), the case law on transformative use and market effect was stacked in Google’s favor. Nonetheless, Judge Chin’s rulings in other cases (e.g. WNET, THIRTEEN v. Aereo, Inc.) suggest that he takes the rights of copyright owners very seriously and that it was essential to persuade him that Google was not merely evading the rights of authors through clever legal or technological structures. The court’s conclusion that the Google Library Project “advance[d] the progress of the arts and sciences, while maintaining respectful consideration for the rights of authors and other creative individuals, and without adversely impacting the rights of copyright holders” pervades all of its more specific analysis.

Data mining, text mining and digital humanities

An entire page of the judgment is devoted to explaining how digitization enables data mining. This discussion relies substantially on the Amicus Brief brief of Digital Humanities and Law Scholars signed by over 100 academics last year.

“Second, in addition to being an important reference tool, Google Books greatly promotes a type of research referred to as “data mining” or “text mining.”  (Br. of Digital Humanities and Law Scholars as Amici Curiae at 1 (Doc. No. 1052)).  Google Books permits humanities scholars to analyze massive amounts of data — the literary record created by a collection of tens of millions of books.  Researchers can examine word frequencies, syntactic patterns, and thematic markers to consider how literary style has changed over time.  …

Using Google Books, for example, researchers can track the frequency of references to the United States as a single entity (“the United States is”) versus references to the United States in the plural (“the United States are”) and how that usage has changed over time.  (Id. at 7).  The ability to determine how often different words or phrases appear in books at different times “can provide insights about fields as diverse as lexicography, the evolution of grammar, collective memory, the adoption of technology, the pursuit of fame, censorship, and historical epidemiology.”  Jean-Baptiste Michel et al., Quantitative Analysis of Culture Using Millions of Digitized Books, 331 Science 176, 176 (2011) (Clancy Decl. Ex. H)” (Authors Guild v. Google, p.9-10)

The court held that Google Books was “[transformative] in the sense that it has transformed books text into data for purposes of substandard research, including data mining and text mining in new areas, thereby opening up new fields of research. Words in books are being used in a way they have not been used before. Google books has created something new in the use of text — the frequency of words and trends in the usage provide substantial information.”

A snippet of new law

Last year, the court in HathiTrust ruled that library digitization for the non-expressive use of text mining and the expressive use of providing access to the visually disabled was fair use. Today’s decision in Authors Guild v. Google supports both of those conclusions; it further holds that the use of snippets of text in search results is also fair use. The court noted that  displaying snippets of text as search results is similar to the display of thumbnail images of photographs as search results and that these snippets may help users locate books and determine whether they may be of interest.

The judgment clarifies something that confuses a lot of people — the difference between “snippet” views on Google books and more extensive document previews. Google has scanned over 20 million library books to create its search engine, mostly without permission. However, Google has agreements with thousands of publishers and authors who authorize it to make far more extensive displays of their works – presumably because these authors and publishers understand that even greater exposure on Google Books will further drive sales.

The court was not convinced that Google Books poses any threat of expressive substitution because, although it is a powerful tool for learning about books individually and collectively, “it is not a tool to be used to read books.”

The Authors Guild had attempted to show that an accumulation of individual snippets could substitute for books, but the court found otherwise: the kind of accumulation of snippets that the plaintiffs were suggesting was both technically infeasible because of certain security measures and, perhaps more importantly, was bizarre and unlikely: “Nor is it likely that someone would take the time and energy to input countless searches to try and get enough snippets to comprise an entire book.  Not only is that not possible as certain pages and snippets are blacklisted, the individual would have to have a copy of the book in his possession already to be able to piece the different snippets together in coherent fashion.”


Today’s decision is an important victory for Google and the entire United States technology sector; it also confirms the recent victory libraries, academics and the visually disabled in Authors Guild v. HathiTrust.

Unless today’s decision is overruled by the Second Circuit or the Supreme Court — something I personally think is very unlikely –, it is now absolutely clear that technical acts of reproduction that facilitate purely non-expressive uses of copyrighted works such as books, manuscripts and webpages do not infringe United States copyright law. This means that copy-reliant technologies including plagiarism detection software, caching, search engines and data mining more generally now stand on solid legal ground in the United States. Copyright law in the majority of other nations does not provide the same kind of flexibility for new technology.

All in all, an excellent result.

* Updated at 4.57pm. The initial draft of this post contained several dictation errors which I will now endeavor to correct. My apologies. Updated at 5.17pm with additional links and minor edits. 





University of Iowa presentation on copyright, mass digitization and the digital humanities

I am giving a talk today on copyright, mass digitization and the digital humanities at the University of Iowa law school. My talk will focus on the ongoing litigation between the Authors Guild and Google and the separate case of Authors Guild v. HathiTrust. The case against Google began in 2005 shortly after Google launched its ambitious library digitization project. The case against the HathiTrust, a digital library that pulls together the resources of a number of American universities, began much later in September 2011.

These cases raise complicated issues about standing, the scope of class actions, statutory interpretation, the interaction of general and specific limitations and exceptions to copyright under the Copyright Act of 1976, and probably a few others besides. However, at the heart of both cases is actually a very simple question — does copying for non-expressive use require the express approval of the copyright owner?

A non-expressive uses one which involve some technical act of copying the above for which the resultant copy is not read by any human being. For example, checking work for plagiarism involves comparing the suspect work against a database of potential sources. It is certainly valuable to know that work A is suspiciously like work B, but that knowledge is entirely independent of the expressive value of either of the underlying works.

Non-expressive use was not a particularly pressing concern before the digital era – from the printing press to the photocopier, the only plausible reason to copy a work was in anticipation on reading it. In the present however, scanning technology, computer processing power and powerful software tools make it possible to crunch the numbers on the written word in all sorts of remarkable ways. The non-expressive use that most people will be familiar with relates to Internet search engines. Search engines direct users to sites of interest based on a complicated set of algorithms, but underlying those algorithms is an extraordinary database describing the contents of billions of individual webpages. To build a database requires copying and indexing billions of individual webpages.

Authors Guild v. Google will determine whether it was legitimate for Google to extend its Internet search model to the off-line world and apply it to paper-based works which had never been digitized. However, the significance of this cases goes well beyond building a better library catalog — although the importance of that should not be casually dismissed — Authors Guild v. Google and Authors Guild v. HathiTrust will shape the future of the digital humanities. If the District Court ruling in HathiTrust stands, as I believe it should, academics who wish to combine data science and a love of literature will not be shackled to the pre-1923 public domain. They will be able to apply the same analytical techniques to the works of William Faulkner as to those of William Shakespeare. More importantly, distant reading empowered by computational analysis will allow scholars to extend their gaze beyond a narrow literary canon or even the few thousand works for most of us can hope to read in our lifetime and address questions on a broader scale.

Slides are available here: Copyright and Mass Digitization, Iowa 2013


Some thoughts on the use of bio photos

I have noticed over the years that whenever someone puts together a bio page for me in relation to a talk or a conference presentation that they tend to grab just any old photo from the Internet. Quite frankly, some of these photographs a more flattering than others. Most of them are not as good as the selfy I took on my iPhone this morning.  Photos from 10 years ago might be considered too flattering in terms of hairline.

Perhaps with some strategic tagging and linking I can get this to the top of the Google search engine.

Photo of Prof. Matthew Sag 2013

Matthew Sag


I also have a full bio page at which contains all sorts of useful information.