As Adam Ziegler writes in “The Harvard Library Innovation Lab,” the mission of the Harvard Law School Library Innovation Lab (LIL) is multidimensional. As LIL says on its website, the lab is composed of “thinkers and doers working at the intersection of libraries, technology, and law.” The LIL projects that we have profiled elsewhere in this issue—the Caselaw Access Project and Perma.cc—represent two of the Lab’s largest-scale undertakings, requiring years of forethought, preparation, and execution. However, because LIL’s mission is to help create an ecosystem around libraries, technology, and law, the Lab is constantly experimenting with new projects—often small in scale or seemingly esoteric in subject matter—under the belief that big insights can be discovered in even the smallest of ideas. Indeed, LIL takes a more experimental tack through its typically smaller-scale projects, called sketches, each intended to shine a light on a particular problem. There is no blueprint or road map for these sketches. Some find new ways of tackling old, persistent problems (CanIFairUse.It). Others coordinate and amplify existing efforts in ways previously infeasible or untested (Local Memory Project). For some, the rollout is somewhat linear where, once the spark of the idea catches, there is a clear short-term goal and a broader vision for potential future applications (Alterspace). For others, while they may begin with certain defined intentions, they might also veer off in varied, unexpected directions as the sketch develops (screenshare).
A mere sketch may contain the seeds of innovation and change.
Below, we share brief commentaries on four of LIL’s sketches written by people who were instrumental in making them happen. We asked each of them a set list of questions (see the breakout box below) to examine what the sketch was originally intended to do, how it came into its current form, and where it is going next. And, in each case, LIL’s exploratory approach to solving niche problems was essential in connecting the dots.
For readers of The Practice, the point is clear: innovation can come from anywhere, and often from unexpected, small places. As you read the following sketches, keep an open mind and think creatively about how extensions of the ideas contained in these examples might impact you—and about what sort of project, no matter how small or seemingly limited in impact, you might want to explore. After all, as the examples below seek to demonstrate, a mere sketch may contain the seeds of innovation and change.
The Survey
We asked representatives of each of the four sketches below the following groups of questions:
Description. What is the project?
Inspiration. What was the spark that got you interested in pursuing this idea?
Evolution. What were the first steps in getting the project off the ground? How was it rolled out? How has the project evolved from its initial conception? Has it led to new uses/avenues that were not part of the original idea?
Collaborations. What other individuals and/or organizations collaborated on this project to help make it happen? What different skill sets and backgrounds were important?
Results. What results have you seen from this project, both proximate (for example, relating directly to the project) as well as more removed (for example, unintended impacts)? How is it impacting the legal profession (if applicable)?
What’s next? How do you see this project—or its impacts—progressing in the short, medium, and long term?
CanIFairUse.It: Crowdsourcing fair use expertise
By Kyle K. Courtney
The project. CanIFairUse.It? (CIFUI) is a Knight News Challenge project that proposed harnessing crowdsourcing technology to help educate communities—especially libraries—about copyright’s fair use doctrine. Often libraries are the center of this type of fair use question, and each year more and more copyright questions are arriving at libraries. The concept of CIFUI was simple: Could a person enter a fair use hypothetical using rapid technology and get experts to answer [YES] or {NO} to real-life fair use questions? While in the past some legal crowdsourcing programs have failed because of a lack of interest among legal professionals to impart knowledge, the CIFUI project generated excitement among potential copyright experts to answer questions as easily as possible.
The average daily reply rate was approximately 80 percent per day. In a monthlong project, that was excellent.
The inspiration. The project was built on a familiar premise: that misunderstandings and myths surrounding fair use can have cultural, artistic, commercial, scholarly, educational, and free-speech implications. Fair use is a doctrine that was designed for works to be usable by anyone, commercial or noncommercial, artist or designer, student or employee. If we fail to use and understand fair use or, more important, fail to educate our patrons and communities, we might be accidentally censoring or preventing innovation, scholarship, and development. The CIFUI project imagined harnessing basic technology to help stamp out these common fair use myths and misunderstandings with outreach and education. There is always a reluctance in the legal community to answer fair use questions beyond the standard “It depends,” but what if there were only two choices? In this way CIFUI could measure the scale and impact of these more challenging fair use questions emerging from our communities. And, with the fair use statute being four factors, there is always a potential for a tie, but we would design the system not to allow a tie vote. In this way fair use is the best model for truly weighing risk.
The rollout. First, we recruited 30 individuals to serve as the experts to test the initial round of staged hypotheticals. Then with the help of the Library Innovation Lab’s Jack Cushman (see “Making the Law Computable”), we picked a basic mass texting program that could deliver the hypotheticals to the 30 copyright experts daily. The program would accept only a [YES] or {NO} reply. This was a way for us to really lean on the experts’ knowledge and ask them, in many ways, how they think a court might rule if it was presented with a particular fair use problem.
Then we created the hypotheticals. Katie Ott (then our copyright fellow, now at Stanford Law) and I took a first pass at writing the scenarios. We designed some easier fair use questions in the beginning of the 30-day pilot as a litmus test. Next, we designed more-difficult questions further into the project. Jack also aided us with the implementation and word choice on these questions, many of which were drawn from real cases, some from pending cases, and some from real-life fair use questions from our community.
The rollout taught us a few things. First, our scenarios were occasionally identified as texting spam because we were sending the same emails to 30 cellphones at once. It was carrier specific. Jack eventually adapted this so that each of the hypotheticals came from a unique phone number. This enabled us to eventually roll these hypotheticals out daily with no technological interference.
The [YES] or {NO}’s were collected anonymously each week, and we could review the results. Later, we also set up an area for feedback about the questions and process. By the third week, some of the copyright experts were sending their own fair use hypotheticals to test on their peers, and we ultimately introduced some of these new fair use scenarios into the project.
Again, the experts themselves were also promised anonymity in their replies. This helped them feel more comfortable about weighing in on the [YES] or {NO} reply.
The team. Including Jack Cushman, Katie Ott, and myself, we had a group of 30 copyright experts from across the United States—including lawyers, librarians, archivists, law professors, deans, and technologists—who participated in the prototype project. We also used real-life questions from Harvard’s Copyright First Responders network.
We were hoping to show that a legal crowdsourcing mechanism could be used successfully.
The results. We had very high participation rates, and very few left the project. The average daily reply rate was approximately 80 percent per day. In a monthlong project, that was excellent. Even though our experts were at conferences, vacations, or work, we still received great responses. Much of that had to do with our simple [YES] or {NO} format, which they could manage easily on their phones each morning we sent them a hypothetical.
The data showed uniform agreement on a few hypotheticals where there was 100 percent [YES] or {NO} answers. For more-difficult scenarios, we had a good range of numbers that still gave a majority, which, we believe, could inform future users about their fair use potential. It was certainly reflective of the areas where fair use and the law have had different outcomes in the courts.
For the legal profession, we were hoping to show that a legal crowdsourcing mechanism could be used successfully. Some crowdsourcing in the past had failed, but the parameters we created—no “it depends” answers, anonymity in answering, ease of voting, and passion for the topic—would help other areas in the law where risk could be weighed by the user, with some expert-level input. Our goal is to provide the public with broad, nondiscriminatory access to legal expertise to make their own decisions.
The future. We are hoping, after one more round of text-based testing with a larger cohort, that we can bring this to scale for library and cultural institutions. Often libraries are the center of this type of question—each year, more copyright questions are arriving at libraries. And the libraries themselves are moving their own copyright expertise in-house, much like the U.S. Copyright Office has done under the leadership of the Library of Congress. And, matching their mission, it is our project’s goal to provide the public with broad, nondiscriminatory access to fair use expertise. We believe that this will not only put the libraries and copyright experts in a great position as experts in their field, but that it will help educate and empower the public to make their own fair use demonetizations to “promote the progress of science and the useful arts.” Imagine a user gets dozens of replies from fair use experts around the United States as a risk-based answer to their fair use question. They could then make their own risk-based determination based on their now-informed use. This influence and education, we believe, would help the public make future fair use decisions more effectively and accurately.
Local Memory Project: A broader platform for local stories
By Alexander C. Nwala
The project. Local and national news media have different priorities. As a result, important local news stories often do not receive enough, or any, national coverage. To learn about a local event, it is crucial to consult the local sources affected. Consequently, local news media is fundamental to journalism.
Before it became a project, LMP was a sketch conceived by Adam Ziegler, who wanted to document the experience of a parade in his hometown of Covington, Louisiana.
The Local Memory Project (LMP) strives to increase the exposure of local news media by providing a suite of tools to help users discover, collect, build, archive, and share collections of stories for important local events from local sources. The first tool, LMP-Geo, discovers local news media organizations that serve a particular location. The second tool, LMP-collection generator browser extension, finds and collects news stories discovered by Geo.
The inspiration. Before it became a project, LMP was a sketch conceived by Adam Ziegler, who wanted to document the experience of a parade in his hometown of Covington, Louisiana. He realized that the news and social media content surfaced by a Google search inadequately approximated his experience from one such parade, the Orpheus parade. The need for a solution to this problem was further refined with the help of Anastasia Aizman into the LMP sketch. The idea for the fundamental LMP function (Geo) was conceived by my Ph.D. supervisor, Dr. Michael Nelson. All I did was design the LMP architecture and implement a simple solution.
The rollout. LMP started off in a different direction, courtesy of my initial solution. Local events that receive national coverage are likely to be important to the affected communities. Therefore, I initially proposed to automatically detect and build collections for such events and subsequently invite users to contribute to the collections. I worked on this solution from late June 2016 to mid-July 2016, until a meeting with Adam and Anastasia helped redirect me to the more practical solution we now call LMP.
LMP-Geo was released on July 28, 2016, and the collection generator browser extension was released in mid-August 2016. Since their initial release, we have added more than 6,000 non-U.S. newspapers from 183 countries and implemented the capability of extracting tweets from Twitter and posts from Facebook. These useful additions were absent from our initial solution.
The team. The success of LMP can be attributed to the collaboration between Harvard LIL and the Web Science and Digital Libraries (WS-DL) research group at Old Dominion University.
From Harvard LIL, Adam, a lawyer and the director of Harvard LIL, conceived the idea for LMP. Anastasia, a developer and designer, advised me on the architectural and aesthetic design of LMP. She was responsible for the addition of “Project” to the initial name of LMP, “Local Memory.” She also designed the LMP logo.
From WS-DL, Dr. Michael Nelson and Dr. Michele Weigle, experts in web preservation and digital libraries, advised me on the architectural design, collection generator, and archiving system of LMP.
Developers ought to be better at sales.
The results. I consider the release of the local news repository one of the most significant contributions of LMP. The repository includes the websites of more than 12,000 newspapers (U.S. and non-U.S.), 1,061 U.S. TV stations, and 2,539 U.S. radio stations. This is a valuable data set to anyone who intends to conduct local news–related research. For example, Social Feed Manager (SFM), an open-source software created at the George Washington University Libraries to help researchers and archivists build social media collections, leveraged the LMP data set to double the number of news outlets’ Twitter accounts in the SFM database.
LMP has also been useful to ordinary users. On September 10, 2017, I collected and shared local news stories from Havana, Cuba, a day after Hurricane Irma impacted the country. A friend of Dr. Weigle’s who had difficulty contacting members of a church in Matanzas, Cuba, reported that she found the local news collection more useful than results from ordinary search.
The future. I strongly believe there is more to come from LMP. There are multiple opportunities to generate derivative projects. For example, one could take the pulse of the United States by conducting topic modeling on U.S. local news sources to identify patterns or topics of interests on a local or national scale. However, we have to do a better job at promoting the LMP tools. Borrowing a phrase from my Ph.D. supervisor, developers ought to be better at sales. It is not sufficient to build a useful tool and expect the community of users to organically become aware of the tool. Therefore, we have to take conscious steps toward publicizing LMP.
Alterspace: A room made just for you
By Clare Stanton and Anastasia Aizman
The project. Alterspace is a new kind of reading room: a simple set of controls lets patrons decide on the color and behavior of lights and sounds in the space, adapting conditions for brainstorming, meditation, or quiet study. On the small scale, it is an installation that debuted at the Cambridge Public Library in early March that allows patrons to enter a space and create their ideal environment for their library visit. On the large scale, it is a replicable project with open-source software that will allow libraries to offer a new type of experience to their patrons.
Typically, patrons don’t have a lot of control over the environment that they are in while visiting a reading room.
The inspiration. We had been talking about an experience in a library, playing with the idea of a location-based sound maze through library stacks. Soon after, we happened on the opportunity to collaborate with metaLAB (at) Harvard and held a brainstorming session at LIL. It turned out our interests were very much in line: we congregated over the idea of creating a Room of Requirement (like in the Harry Potter series), one that, by utilizing various levers, would morph into whatever space a library patron desired. Personally speaking, though I know I share this with others on the team, I was very motivated by the idea of giving patrons agency over a physical space. The library is a fantastical place already where anyone has access to physical and educational resources—Want to learn a second language? Borrow a book or a laptop?—but it still puts constraints on the visitors’ control over the space that they occupy. For many patrons in a library, from the indigent to children, that is not an option in their lives.
The rollout. Initially this project came out of an exciting opportunity to work with our friends at metaLAB—a group at Harvard focused on research and experimentation in the arts, media, and humanities. Our teams were immediately interested in exploring the room of requirement idea that Anastasia describes above. Typically, patrons don’t have a lot of control over the environment that they are in while visiting a reading room, so Alterspace became a project to try to give that autonomy and control to library users. Our first installation was a prototype to gather feedback to iterate on—our eventual hope is for the project to be replicable by any library that would like to provide it to their patrons.
The team. metaLAB’s Matthew Battles and Jessica Yurkofsky were our partners in crime with their design and installation expertise. Keith Hartwig constructed a room inside the Cambridge Public Library’s Rossi Room for our initial testing installation. Alisa Kolot researched and either sourced and edited or created the sounds that we’re using. Andy Silva of LIL and Sam Stites of Sentenai helped solve a mind-boggling networking problem, without whom our first go in the Cambridge Public Library wouldn’t have been a possibility. Support from Library Innovation Lab, the Harvard Law School Library, and the Knight Foundation was imperative to making this happen.
The results. Even before the first run of this went live, we had already seen in our discussions with our partner libraries—both the Cambridge and Somerville public libraries will be hosting Alterspace in the coming months—that patrons might be looking for this type of experience. From meditating in private study rooms to gathering in more-open areas to collaborate with friends, the public sees the library as a place they can adapt.
We hope that by providing a tested, open-source tool kit we can make Alterspace easily reproducible.
The feedback we received at the Cambridge Public Library was positive. Both children and adults were delighted by the space and spent time reading or relaxing inside. We have heard from the library that older patrons were especially happy to have programming for adults that was fun and even playful. It sounds like they plan to bring that feedback forward into other events they offer.
We’re not sure if this will impact the legal profession just yet, but with our open-source software, who knows where Alterspaces could pop up—does your law firm need a room of requirement to help attorneys focus, meditate, or litigate?
The future. We hope that by providing a tested, open-source tool kit—this project is hosted on GitHub—we can make Alterspace easily reproducible. We’re doing quite a bit of testing of technology, user experience, and testing this project in library spaces like the Cambridge Public Library, Somerville Public Library, and Langdell Hall at Harvard Law School. We hope that anyone else in a public library, or any other public or private setting, would be able to get Alterspace quickly up and running.
Screenshare: The unintentional multitool
By Ben Steinberg
The project. Screenshare is a tool for posting images to a Slack channel and displaying them on a monitor in our shared office, called LILspace; screen-api, a similar application, lets you post text to a web API and displays it on another monitor. In each case, there’s a remote server, and the local component runs on a Raspberry Pi, a hobbyist’s single-board computer.
The development process was a quick, one-person job, probably done as a fun Friday project.
The inspiration. Matt Phillips’s original motivation for writing screenshare was to make a toy—something to liven up the office. Jack Cushman developed the screen-api project as a teaching tool for his course, Programming for Lawyers (see the breakout box in “Making the Law Computable”), where the idea was to give the students a simple way to affect the outside world using code in the very first class.
The rollout. In either case, the development process was a quick, one-person job, probably done as a fun Friday project. (LIL has a practice, possibly honored more in the breach than the observance, of fun Fridays, a day a week for working on experiments and side projects.) Initial deployment of screenshare was improvisational, using equipment lying around the office and LIL’s existing web server. We got more organized for subsequent deployments, using the configuration management and orchestration system SaltStack to deploy code and settings to the Pi and a virtual server in the cloud. The deployment of screen-api benefited from that organization. The screenshare code itself was completely rewritten last summer to use Django Channels, a system for communication using WebSockets.
Screenshare still functions as a toy, but it’s an important element of our office culture: we post images that are amusing, pertinent, instructive, picturesque, or absurd. It also has a side use as an impromptu slide projector; we’ve had people give talks without slides, where the audience can post relevant images at will. The screen-api is in regular use not as a teaching tool but as a kind of dashboard where we can automate the posting of arbitrary information: for instance, I am presently exposing the time and domain of the most recent Perma capture (see “Pausing the Internet”) and the progress of a long-running batch job for compressing Caselaw Access Project images (see “Making the Law Computable”).
The team. As mentioned, two LIL developers created these systems. Matt Phillips wrote screenshare initially. Jack Cushman wrote screen-api and rewrote screenshare last summer. My work on these systems has included deployment using SaltStack, troubleshooting, and a few code updates. All three of us are developers; Jack is also a lawyer, and I’m a librarian, though I’m not sure those skill sets were critical to the projects.
Screenshare demonstrates the utility of small, “inconsequential” projects for organizational learning and reuse.
The results. Although it wasn’t intended, screenshare has turned out to be an important setting for technical experimentation and improvement, and the screen-api has proven useful for monitoring infrastructure. Without discounting the merits of both projects as elements of workplace culture, as art, and as toys or teaching tools, they have great instrumental value as nonessential, low-stakes contexts for trying new technologies. By the same token, they’re a place where we can see the value of throwing a project away and starting over—something that’s nearly impossible to contemplate for a large and consequential piece of software.
The future. I think both of these projects will change incrementally, and possibly find new uses in LIL, in the library environment or the classroom. More generally, I think they demonstrate the utility of small, “inconsequential” projects for organizational learning and reuse. As LIL continues to develop large projects like Perma.cc, H2O, and the Caselaw Access Project, we will make room for the small and secretly consequential.
