Wired Campus icon


Internet Seen as Leveling Opportunities for Scientists


New Editing Process Seeks to Improve Wikipedia's Accuracy

August 25, 2009, 10:00 AM ET

Labeling Library Archives Is a Game at Dartmouth College

Professor Mary Flanagan wants students to go online and label library archives – for free.

Ms. Flanagan, a digital-humanities professor at Dartmouth College, is creating an Internet-based game in which users create descriptive tags for library images to improve searching through the library's database. Although the program will be tested at the college’s library, Ms. Flanagan says the game will be open source and available for others to download and build upon.

She says the program could save libraries time and money. “It’s costly and time consuming to go in and add keywords,” she says. “If you create a game where people actually are actually getting points for generating metadata, you create a system of motivation and a fun way of doing this kind of stuff that people, it turns out, will do for free.”

The two-dimensional game will show an image, and players will have to enter words that describe it. As players race the clock, they will receive points when their phrase matches another person’s.

“Knowing the Dartmouth folks, we’re going to get some really in-depth research as the game progresses,” Ms. Flanagan says. “It will bring out a competitive edge to knowledge.”

The project received $49,015 from the National Endowment for the Humanities and is expected to be completed by next summer.


1. kphagen - August 25, 2009 at 04:42 pm

Whitewashing the fence has just gone digital!

2. mschlat - August 25, 2009 at 04:42 pm

Isn't this just the same idea as Google Image Labeler (http://images.google.com/imagelabeler/)? Not that I don't think this is a good idea for private archives, but how is this different from existing programs?

3. jmmason - August 25, 2009 at 05:04 pm

Whitewashing the fence has been digital for years now. Do you think Wikipedia authors get paid? Most Amazon book reviewers (some do get free books from publishers)? (Self-serving note alert: I wrote a paper in 2007 called "Tom Sawyer Production on the Internet: Getting the Good Stuff In, Keeping the Bad Stuff Out" -- I've presented it several times, but haven't quite finished the theoretical model to my satisfaction so haven't tried to publish it yet.)

Like mschalt, I wonder how this differs from the "ESP Game" (a scholarly research project by Dabbish and von Ahn at Carnegie-Mellon, later licensed by Google as its "Image Labeler").

4. janawoo - August 25, 2009 at 06:36 pm

My library's archival photo database has an "add your tag" feature, and though it's not a game, it didn't cost anywhere near $50,000 to implement --I think it took me about 20 minutes to set it up.

In my opinion, you still need a trained information professional to review the metadata for accuracy and consistency, because as we know, pictorial meaning is notoriously slippery. The user input certainly does help though.

5. amandasc - August 26, 2009 at 10:18 am

I played this exact same game when I was in grad school for Library and Information Science. Wish I had my notes so I could link to it, but this idea is not so original.

6. maryflanaganify - September 07, 2009 at 02:34 pm

Hi everyone, thanks for the interest in the project!

mschlat and jmmason, great points. Yes we are very inspired by the work of Luis von Ahn and efforts at Google. Our work builds on some of this prior work but adds new types of scale, game motivations and interactions that are not described in this short blog post. Personally, this project emerges from a longstanding interest in alternative search engines and visualization tools I have developed in the past.

It is important for us to give nonprofit institutions the ability to use crowdsourcing technologies if they so choose while still being able to control the collections as they are bound by various rights and responsibilities. Additionally, our play innovations by professional game designers can be shared among developers of other archival projects, and the application used beyond image tagging to larger knowledge frameworks. More to come as we progress! Keep your eyes on http://www.tiltfactor.org for more information on this project and on our other work.

janawoo - One of the purposes for the funding is to analyse the validity of the process for both the info sciences and game development community through formal assessment. We'll share our results as our research is completed. Verification is a genuine challenge for crowdsourcing, but we have some good ideas for how assessment can shift the design of crowdsourcing systems.

And amandasc, I hope you will find as we progress that the originality comes from the nuances in the design. The actual gameplay is still in development -- the project was just announced, and game design is an iterative process -- so there is much more to come. We're developing the social networking components of the sharing system. Overall, our team is committed to making interesting hybrid tool-games that, from an open source perspective, further civic engagement and foster participatory, player generated contributions to knowledge from all social spectrums.

Thanks for your feedback and I look forward to hearing from you!

Add Your Comment

Commenting is closed.