Column Editor: Will Cross (Director, Copyright & Digital Scholarship Center, NC State University Libraries) ORCID: 0000-0003-1287-1156
QUESTION: A publisher asks, “Now that the recent Supreme Court term is finished, what ended up happening in the Google v. Oracle case?”
ANSWER: This was a particularly busy Supreme Court term for copyright issues, with major decisions (discussed in earlier Q&A columns) about access to the law and sovereign immunity. One case that many copyright experts were watching closely, however, was not decided this term — Google v. Oracle.
The issue in Google v. Oracle is whether the technology corporation Oracle can claim a copyright in Java APIs (“application programming interface,” specifications that allow programs to communicate with each other) and, if so, whether Google infringed these copyrights when they used the same names, organization, and functionality as the Java APIs in their software.
When originally considered in 2012. Judge William Alsup of the Northern District of California ruled that APIs are not subject to copyright because “there is only one way to declare a given method functionality, [so that] everyone using that function must write that specific line of code in the same way.” Oracle appealed the ruling to the U.S. Court of Appeals for the Federal Circuit and in 2014, the Federal Circuit found that the Java APIs are copyrightable but left open the possibility that Google might have a fair use defense.
After some procedural back and forth, the case returned to the district court for a jury trial on the fair use issue and a jury unanimously agreed that Google’s use of the Java APIs was fair use. In 2018, however, the Federal Circuit took the unusual step of rejecting the jury’s verdict and held that Google’s use was not a fair use as a matter of law.
Google appealed to the Supreme Court, which was expected to hear oral arguments in March of 2020. Unfortunately, oral arguments were postponed in light of concerns surrounding COVID-19, and Google v. Oracle was one of several cases from the 2019-20 term to be postponed until next year. As of this writing, arguments are slated for October 7, 2020. Whenever the case is heard and decided, it raises significant questions about technical interoperability and about fair use. Librarians, researchers, and publishers are watching this case closely since it will have a profound impact on the way the internet functions and the way the law understands use of protocols and procedures in light of the relationship between ideas and expression.
QUESTION: A collections librarian asks, “What is happening now with the Internet Archive’s Emergency Public Library?”
ANSWER: As discussed in previous Q&A columns, the Internet Archive’s Emergency Public Library (EPL) was offered as “a temporary collection of books that supported emergency remote teaching, research activities, independent scholarship, and intellectual stimulation while universities, schools, training centers, and libraries were closed due to COVID-19.” Functionally, the Internet Archive removed their waiting lists for roughly 1.4 million in-copyright books in their collection, making them immediately available to anyone on the Internet. The program was launched in March of 2020 and intended to run through June 30 “or the end of the U.S. national emergency, whichever is later.”
In early June four publishers — Hachette, Penguin Random House, Wiley, and HarperCollins — sued the Internet Archive for “copyright infringement on a massive scale,” seeking unspecified damages and attorney’s fees. In response, the Internet Archive announced that they were ending the program ahead of the scheduled June 30th closing date and returning to their standard practices including “controlled digital lending” (CDL).
With the EPL closed, this controlled digital lending practice is expected to be the focus of the case going forward. CDL is described as “an emerging method that allows libraries to loan print books to digital patrons in a ‘lend like print’ fashion.” By promoting an “owned to loaned” ratio where a library circulates the exact number of copies of a specific title it owns, CDL offers one path to bring library lending into the digital environment. Proponents of CDL are hopeful that it can help libraries move beyond the licensing culture that has dominated online access in the absence of a statutory equivalent to the first sale doctrine embodied in Section 109 of the Copyright Act.
While CDL has been in use for several years by the Internet Archive and others, the need for a way to share digitally has become increasingly urgent as the COVID-19 pandemic is preventing many libraries from engaging in physical resource sharing. The Project ReShare Community, a group of libraries, consortia, information organizations, and developers, plans to develop a minimum viable product to support CDL in the fall, and many libraries are exploring CDL for their own use in the fall and spring. With CDL as an emerging model, this lawsuit may be a critical first test for the fair use analysis. You can read more about CDL on this website: https://controlleddigitallending.org/.
QUESTION: A media studies professor asks, “With many students completing their work online this semester, we’re exploring online media platforms for gathering and sharing materials. I know the basics about copyright, open licensing, etc. but I’m wondering how algorithms used on those sites impact how content is shared.”
ANSWER: This is another question that demonstrates the way that copyright issues require us to think about questions that go well beyond black letter copyright law. Much has been written about YouTube’s Content ID service, a digital fingerprinting system developed by Google which is used to easily identify and manage copyrighted content. Moderation at scale is notoriously difficult and automated systems such as Content ID as well as similar systems at social media sites like Twitter and Instagram particularly struggle to with fact-sensitive equitable exceptions such as fair use. Many of these systems have a history of both over- and under-enforcing copyright claims by leaving up infringing content while at the same time removing content that would clearly be permitted under fair use.
In particular, there is a robust history of misguided and predatory takedown notices such as the notorious lawsuit resulting from a parent, Stephane Lenz, posting a 29-second clip of her 13-month-old son dancing to Prince’s song “Let’s Go Crazy” (Lenz v. Universal Music Corp., 801 F.3d 1126 (9th Cir. 2015) or Soundcloud’s removal of a totally silent “remix” of experimental artists John Cage’s “4’33,” a piece to be (not) played in three movements of total silence.
In the past few months, we have seen both silly and very serious examples of this overreach. On the lighter side, an artificial intelligence project based on a famous parodist and focused on machine learning was recently flagged on Twitter. Mark Riedl, a researcher at Georgia Tech created a machine learning model that creates lyrics that match the rhyme and syllables schemes of existing songs. He calls it “Weird A.I. Yancovic” as an homage to the parodist/national treasure “Weird Al” Yankovic, who is best known for creating his own parodies of famous songs.
Riedl posted a video to Twitter in May of 2020 with AI-generated lyrics and the instrumental part of Michael Jackson’s “Beat It.” Twitter quickly took the video down after the International Federation of the Phonographic Industry, a coalition of some of the record industry’s biggest companies, submitted a copyright takedown notice.
Traditional parodies such as those by “Weird Al” the singer are certainly permitted under fair use. Although Yankovic often seeks permission for personal ethical reasons, the leading Supreme Court case on fair use Campbell v. Acuff Rose explicitly addressed these issues and found that commercial parody by the rap group 2 Live Crew was fair. Riedel argues that, like 2 Live Crew and “Weird Al,” songs created by his Weird A.I. should be permitted under fair use. While the legal issues are relatively clear — a court would likely spend more time wrestling with questions about authorship of the AI-generated lyrics than the fair use analysis — Twitter’s takedown policy removed the materials anyway.
In a more serious example from YouTube, videos of police abuse and Black Lives Matter protests have been regularly taken down based on copyright claims. As documented by the Lumen Database (https://www.lumendatabase.org/blog_entries/background-audio-takedowns), most of these cases are “false positives” where the copyright claim is based on background music, such as Marvin Gaye’s “Let’s Get it On” or 2Pac’s “Keep Ya Head Up,” playing in the public spaces where protests are occurring, rather than as a core part of video itself.
In the context of higher education, some faculty instructors have even proposed taking this a step further by deliberately including copyrighted content in their lecture videos at volume levels too low to distract a human viewer but present enough to trigger the copyright bots. At your own institution, the best approach is to be aware of and thoughtful about these issues and to offer guidance and training where appropriate. Students use platforms like these all the time and the fall semester offers many opportunities to use these platforms for coursework, projects, and scholarship. A clearer understanding of the rules and dangers will be invaluable for your own work as well as for the work of instructors, creators, activists, and anyone seeking to communicate online.
QUESTION: A history professor asks, “As we’re moving into online instruction I see a lot of Zoom backgrounds that include popular media. Are there any copyright concerns for me or my students?”
ANSWER: As many of us have moved to meetings online it is increasingly common to see digital backgrounds pop up behind people’s faces. Some of these are personal photographs or openly licensed images, often of our own campuses or workplaces. Just as often, however, the background uses copyrighted materials to create the illusion that the attendee is speaking from a different location such as the Simpsons family couch or the bridge of the Starship Enterprise.
While these images are often protected by copyright, using a still from a film or TV show in this way is almost certainly permitted under fair use. Personal, noncommercial use of these small portions is arguably transformative and unlikely to cause substantial market harm. The only areas of concern might be extreme cases such as commercialization of these images, for example by setting up a database that sells images from popular media, or using an unlicensed stock photo of a fabulous location.
In contrast to the relatively low legal risk of using these backgrounds, there is substantial value. These backgrounds offer some whimsical and expressive value for individuals who want to personalize their experience and lighten up what might be a formal setting to build community in a socially distanced course or meeting. The real value in these backgrounds, however, comes from the purpose they were designed for: giving users (especially but not only students) a tool for masking their physical surroundings.
As online education requires us to invite our peers, our supervisors, and so many others into our most intimate spaces, it is critical that we all have some way to retain some degree of privacy. This is especially true for students living with extended family, working from a car parked near public wifi, or otherwise connecting from a space they might not be comfortable sharing with a large group of relative strangers. As is so often the case, an overly restrictive view of copyright can mask the core values of education at the expense of those who need privacy, autonomy, and agency the most.