Future Tense

The Crowdsourcing of Talent

Scientists are using video games to make major breakthroughs. Are they revolutionizing how we manage labor at the same time?

Older woman typing on a laptop.
Crowdsourcing has ushered in dramatic changes to fields like graphic design and corporate R&D

Photograph by Johnny Greig/iStockphoto.

This article arises from Future Tensea collaboration among Arizona State University, the New America Foundation, and Slate. On Feb. 29, Future Tense will host an event on the Make movement and do-it-yourself innovation in Washington, D.C. For more information and to sign up for the event, please visit the NAF website.

In the summer of 2007, I went to Calgary, Canada, to report on a startup. It was a heady time: Venture capital was plentiful, and the company, Cambrian House, was generating a lot of excitement. The previous year I’d written an article in Wired called “The Rise of Crowdsourcing,” which proclaimed the dawn of a new era: The many would soon do the work of the few; the crowd would display its vast wisdom in every area of human endeavor; the community would replace the company.

In the wake of the article, crowdsourcing became a buzzword and a flood of specious business plans started arriving in my inbox. Cambrian House seemed different. CEO Michael Sikorsky had built a large community of some 5,000 would-be entrepreneurs. Ideas for new software programs were suggested, discussed, then carefully culled through a rigorous voting process. People would cheerfully volunteer to work on the winning projects. Designers would design; coders would code; the biz dev folks would do whatever the hell biz dev people do. They would all be compensated in royalty points, IOUs that could be cashed in once the dough started rolling.

It was a radical rethinking of one of capitalism’s central assumptions—that labor is best organized from the top down, which is to say, by management. “It’s like the dark ages before Newton, when we mistook physics for magic,” Sikorsky told me. “We know there’s magic in crowdsourcing. We know it works. We just don’t know how it works.”

Except that ultimately, it didn’t work. A year after my visit, Cambrian House had failed to bring a single product to market. Crowds might self-organize, it seemed, but they didn’t necessarily do it well. (The company found later success by reinventing itself as Chaordix, which essentially provides a backend computer system for crowdsourcing projects at large companies.)

But the idea of totally reorganizing a labor market from the bottom up remained elusive. Crowdsourcing has ushered in dramatic changes to fields as disparate as graphic design, journalism, and corporate R&D, but it has yet to fundamentally challenge the top-down, command-and-control paradigm of how most of the world does business. Instead, crowdsourcing has become a best practice to accomplish discrete, often rudimentary tasks, such as designing logos or scouring government documents for scandalous malfeasance.

Recently, though, I’ve begun to see glimmers of that old magic in some novel experiments that occupy the nexus where crowdsourcing, games, and big data all meet.

Last year, the prestigious journal Nature Structural and Molecular Biology published an article that revealed the structure of an enzyme used by retroviruses similar to HIV. The achievement was widely viewed as a breakthrough. Who solved the riddle? A bunch of video gamers. Foldit, a novel experiment created by a group of scientists and game designers at the University of Washington, had asked the gamers—some still in middle school and few boasting a background in the sciences, much less microbiology—to determine the how proteins would fold in the enzyme. Within hours, thousands of people were both competing against (and collaborating with) one another. After three weeks, they had succeeded where the microbiologists and the computers had failed. “This is the first example I know of game players solving a long-standing scientific problem,” David Baker, a Foldit co-creator, wrote at the time.

It wasn’t to be the last. Foldit is humming along nicely and in January revealed an accurate model for another highly complex enzyme. Another of Foldit’s co-creators, Adrien Treuille, has gone on to start a similar game, eteRNA, in which gamers create designs for synthetic RNA. Every week eteRNA’s scientists actually create the top-scoring designs in the test tube. “When we can predict protein folding, we’re going to be able to build the equivalent of the airplane inside your body, and do amazing things.”

Foldit, eteRNA, and their ilk may well revolutionize how we treat disease. But they’re holding out another promise as well: a realization that conventional management practice is often dead wrong about who is best suited for a task. The best way to match talent to task, at least in the world of nanobiotechnology, isn’t to assign the fanciest degrees to the toughest jobs, but rather to observe the behavior of thousands of people and identify those who show the greatest aptitude for the cognitive skills that task requires.

“You’d think a Ph.D. in biochemistry would be very good at designing protein molecules,” says Zoran Popović, the University of Washington game designer behind Foldit. Not so. “Biochemists are good at other things. But Foldit requires a narrow, deeper expertise.”

Or as it turns out, more than one. Some gamers have a preternatural ability to recognize patterns, an innate form of spatial reasoning most of us lack. Others—often “grandmothers without a high school education,” says Popovic—exercise a particular social skill. “They’re good at getting people unstuck. They get them to approach the problem differently.” What big pharmaceutical company would have anticipated the need to hire uneducated grandmothers? (I know a few, if Eli Lilly HR is thinking of rejiggering its recruitment strategy.)

In a January speech, Treuille noted that he and his colleagues at eteRNA were able to “filter through hundreds of thousands of people who are experts at very esoteric tasks.” They are able, in other words, to match talent to task with exceptional efficiency. Not based on someone’s CV, and not based on the magic of “self-selection,” but rather through the thousands of data points generated by the gameplay.

The success of Foldit and Treuille’s eteRNA, then, depends on a far more sophisticated use of crowdsourcing than anyone was envisioning as recently as 2008, when I published my book on the subject. “In the future,” Treuille said in his speech, “I can imagine as grand challenges emerge we can come up with games and puzzles that essentially exploit the skills required and find people who are experts at these kinds of problems and the person who owns this network owns something very valuable.”

Popovich and his UW colleagues are at work on a new game in which people design motors that work at the molecular level. “We’re using nature’s tools to create designs that didn’t exist in nature for purposes of effectively fighting new diseases.” Currently, he says, there are 50 people, in the world, qualified to do this work. “I can increase that by two orders of magnitude,” Popovich says. “I can get to 5,000 people really easily.”