The Web

Putting the IT in Crowdsourcing

Can IT benefit from mass collaboration? Consider these quick-start ideas for using the wisdom and talents of the crowd for technology operations.

WHEN THE UNIVERSITY OF UTAH wanted a new bus station designed, it turned to the masses for input. The idea was inspired by a paper written by Daren Brabham, a PhD candidate and graduate teaching fellow at the university, who positioned crowdsourcing as a way to increase public participation in public planning. Tom Sanchez, who had just moved to the university to head up its urban planning program, took Brabham’s paper and turned it into the rationale for a grant from the Federal Transit Administration. The Utah Transit Authority would work with the university in implementing the design project.

That grant turned into Next Stop Design, a web-based competition from June through September 2009, in which the general public submitted bus stop designs and vetted the submissions to determine the best ones. Along the way, participants from 127 countries submitted 260 designs and cast over 11,000 votes. Interestingly, fewer than 8 percent of site visitors were from Utah, and only half of those came from Salt Lake City, where the bus stop is located. The three winning entries came from Greece, India, and South Dakota.

“The goal was to increase public participation in a transit- planning effort,” says Brabham, who acted as the project manager and is completing his dissertation on the topic of crowdsourcing and public participation. “We consider it a success—thousands of voices in a democratic process. Most people were frequent riders of buses in their areas. This was a user community, which speaks to a lot of user innovations—that users know best and have a lot to contribute.”

Crowdsourcing in Action

According to Jeff Howe, author of the 2006 Wired article that coined the term, “crowdsourcing is the act of taking a job traditionally performed by a designated agent and outsourcing it to an undefined, generally large group of people in the form of an open call.” The idea is to put the labors of many people to work. The bus stop project, for example, would formerly have been handled by an architectural firm hired by the transit agency.

Brabham adds another layer of specificity to the definition. He believes there’s a traditional management aspect to crowdsourcing. “I define it as an online distributed problem-solving production model,” he says. “Somebody in the organization needs something done and asks the crowd to do it. It isn’t organic. The crowd does not decide what it does next. It’s a very top-down thing, blended with an open, creative process.”

The advantages to crowdsourcing are many. Problems can be addressed with relatively little expense and often with great speed; the organization can gain access to skills and talents outside its immediate circle; and often the crowd can turn out to be smarter or wiser than any one individual in it.

Examples of crowdsourcing in higher education can be found across disciplines:

  • The Amazon Mechanical Turk, which calls itself “a marketplace for work that requires human intelligence,” has been tapped by universities for crowdsourcing projects. The service, from Amazon.com, allows users to break up a project into discrete units—called human intelligence tasks (HITs)—and outsource them en masse to workers for compensation that frequently is just a few cents per HIT. In 2008, Teesside University in the United Kingdom turned to the Mechanical Turk when it needed help completing a project to create online skills development maps for some of its courses. And last month, the Computer Vision Lab at the University of California-Irvine used the service to get hundreds of people to view video sequences as part of a research project on human motion.
  • Dartmouth College (NH) Professor Mary Flanagan won a $50,000 grant from the National Endowment for the Humanities to develop a set of games that can be used by the public to add keywords and descriptive data to specialized libraries and archival databases. Her current focus is eliciting metadata for collections that may—for copyright reasons—only be viewed by a single person at a time or that would benefit from expert input.
  • In December 2009, when the Defense Advanced Research Projects Agency (DARPA) positioned red weather balloons in 10 cities around the country as part of a competition to explore how large problems can be tackled using social networking, MIT used crowdsourcing to win the $40,000 prize. The MIT team obtained the precise global coordinates of the balloons in less than nine hours by encouraging participants to register on a special website set up for the contest and to invite others to do so as well. Anybody who identified the coordinates of a balloon, or somehow influenced the registration of somebody who did so, was promised some portion of the prize money. By the start of the contest, MIT had 5,000 people in its network.

So Where’s the IT in Crowdsourcing?

Brabham, who is joining the University of North Carolina- Chapel Hill in May as an assistant professor, believes the application of crowdsourcing can lend a tremendous amount of knowledge to the right kinds of endeavors. “There’s so much flexibility for how it can be used,” he insists. “You can solve anything if you just frame it right, and build a community around it.”

That is precisely why he is surprised that higher ed IT departments have not caught on to the value of crowdsourcing for their own work. “All [crowdsourcing] is is a website that distributes tasks, aggregates solutions, and allows for ranking mechanisms for people to vote on the best things. That’s really at the heart of everything,” he marvels. In the spirit of distributed problem solving, Brabham offers some suggestions for quick and easy ways to use crowdsourcing to extend and improve IT operations:

How-to videos. Crowdsource the creation of videos explaining how people can do various things with their computer or network connections. “This could be crowdsourced to students on campus or beyond,” Brabham notes. All it would require, he says, is making the video screen-capture software available through a campus license and letting the campus technology center check out equipment such as microphones.

Video transcription. Crowdsource the transcription of videos for people with hearing disabilities or into other languages for students whose first language isn’t English.

Computer lab upkeep. Want to job out the work of checking computer labs for problems, cleanliness, or overall usage patterns? “Task people with duties, such as dropping by certain computer labs at certain times and taking a quick note of how many students are using them, if they’re clean, if they have error messages,” suggests Brabham.

Distributed tech support. Encourage students and faculty to participate in an online “faculty” of technical expertise, he says. When somebody needs tech support, the system can connect that user to the appropriate person, whether he or she is across campus or down the hall.

Problem reporting/beta testing. Get people to report where the wireless gaps are in that pervasive-wireless rollout. Tap into your campus community to test your web, mobile, gaming, and desktop applications during development.

Market research. Use a prediction market platform such as Inkling (which polls participants via an online-trading mechanism) to gauge the value proposition of a new program in your institution. As you develop a new IT policy, get input from the campus community to mold it in ways that make sense to them. “From an administrative standpoint, it may help you predict or alleviate tensions for putting out a policy that nobody likes,” Brabham points out.

Community problem-solving. Post your biggest technology challenges to the campus community—or beyond: For instance, “‘We have a new building going in that’s going to suck too much power off the grid. How should we allocate resources?’ There is an empirically right answer,” says Brabham. “So you cast a wide net hoping to find that needle in the haystack.”

Ultimately, crowdsourcing is “all about being creative in the way you frame a question and knowing what will motivate your crowd,” Brabham concludes, and then waxes a little lyrical. “Maybe these types of things can improve the world somehow with all sorts of ideas—to get people involved in government, to solve tough diseases that drug companies don’t want to research because they’re so rare. That’s my grand vision—to aggregate intellect.”

12 CROWDSOURCING GUIDELINES

DAREN BRABHAM, a crowdsourcing researcher at the University of Utah, points out that to attract a crowd, you don’t necessarily have to offer cash as a reward. It could be inkind meal-ticket money, sports tickets, or other on-campus spiffs that don’t cost the institution anything. Plus, he adds, “You’d get a lot more out of it in PR and writing about it in the alumni magazine.” Here are his other rules of thumb in setting up a crowdsourced project.

  1. Make sure the tasks are small, yet work toward a clear goal that’s big and for the common good.
  2. Use social networking tools to succeed.
  3. Where the reliability of information is vital to the mission, figure out ways to minimize fakery.
  4. Let the crowd clean up your data for you.
  5. Listen to people and respond to their advice—but retain final ownership over the ultimate decision.
  6. Give recognition to top users of your crowdsourcing efforts by points, contributions, and ranking lists.
  7. Make it fun and don’t bog it down with too many rules. That said, make sure you have clarity, so that people know the one or two things they absolutely must do to contribute.
  8. Expect your contributions to come en masse—so the servers better be able to handle the load.
  9. Consider giving the community a chance to interact with the paid staff as part of the reward of participation.
  10. Expect “super-volunteers” to do the bulk of the work and pay attention to what they tell you.
  11. Use paid staff to set up the operations. But use volunteers to manage the work.
  12. Make sure that whatever you want done is of interest to people.
comments powered by Disqus