Gaming the Archives

There’s no shortage of fabulous archival material lurking in college and university collections. The trick is finding it.

Without good metadata—labels that tell researchers and search engines what’s in a photograph, say—those archives are as good as closed to many students and scholars. But many institutions don’t have the resources or manpower to tag their archives thoroughly.

Enter Metadata Games, an experiment in harnessing the power of the crowd to create archival metadata. A team of designers at Dartmouth College, working with archivists there, has created game interfaces that invite players to tag images, either playing alone or with a partner (sometimes a human, sometimes a computer). Solo players think up tags to describe the images they see; in the two-player scenario, partners try to come up with the same tag or tags.

Metadata Games is the brainchild of Mary Flanagan, an artist-designer who’s a professor of digital humanities at Dartmouth. She also directs the Tiltfactor laboratory, which is dedicated to exploring “critical play—a method of using games and play to investigate issues and ideas.” Talking to Peter Carini, the Dartmouth college archivist, about the challenges of tagging special collections got Ms. Flanagan thinking about whether an open-source game interface could help solve the problem.

Mr. Carini was enthusiastic about the idea. “We have several hundred thousand images in our collection, and like a lot of archives and special collections, there’s not a lot of intellectual access to those materials,” he says. “We need better access, but we don’t have the staff to be able to produce that access.”

To test the game interfaces they designed, the Tiltfactor team selected about 200 images from the college’s archives. About 70 percent of them belonged to a collection of historic Arctic images, while the rest were contemporary images. Then the designers invited a small group of players to take part in a small pilot project.

The pilot phase, which just wrapped up, yielded promising results. The players generated 6,250 tags, according to Ms. Flanagan—about 32 or 33 per image. “I would say that over 90 percent of the metadata was useful,” she says. She and Mr. Carini compare that to a Library of Congress experiment collecting metadata from the public on images in its Flickr collection; about 80 percent of those tags were deemed useful, Ms. Flanagan says.

One thing MetaData Games has going for it?  “It’s a lot of fun,” says Mr. Carini, who played some of the games with his son. Ms. Flanagan has thought a lot about the motivational aspects of gaming. “Games are becoming more and more part of what people want to do,” she says. Much of her work revolves around the idea that “what you’re doing in games matters. Games are meaning-making machines.”

Mr. Carini and Ms. Flanagan emphasize that archivists should be in charge of the process, to check the quality of the metadata being collected and filter out skewed results if necessary—and to reassure the archival community that the results will meet its standards. But the time- and labor savings could still be substantial.

Mr. Carini thinks that overworked archivists will be receptive, especially if the game interfaces are able to work well with search engines, and if some of the work of vetting the results could be handled by computers. When he and Ms. Flanagan gave a presentation about Metadata Games at the most recent New England Archivists’ meeting, “there was enormous interest in it,” he says. “I think people would be very excited to have a tool like this.”

For the first phase of the project, Ms. Flanagan got a $50,000 start-up grant from the National Endowment for the Humanities’ Office of Digital Humanities for the project, along with a fellowship from the American Council of Learned Societies. She and Mr. Carini will be further analyzing and sharing the results of the pilot project. They’ll need to find money to support the next round of development, and Ms. Flanagan hopes to find institutional collaborators.  “There’s no reason this couldn’t hook up to other kinds of systems,” like the Omeka content-management system created at George Mason University, she says.

“It’s an experiment,” she says. “And I think the experiment works.”

Return to Top