Big-name companies such as PepsiCo and Kraft Foods Group have run contests to crowdsource new product ideas. Topcoder, an open innovation platform that awards prize money for software development and design projects, has attracted hundreds of thousands of participants. Now a start-up called Driven Data is working to enable online competitions that give data scientists a chance to vie for cash, glory—and social good.
Driven Data, a for-profit housed at the Harvard Innovation Lab, develops those competitions working with mission-driven clients. “We want to help them frame a question that they really care about—something hard and hairy,” says Greg Lipstein, cofounder. Clients pay for that service and also put up the prize money; Driven Data runs the contest.
The first competition, launched this past October, aims to help school districts spend their money more strategically. Education Resource Strategies (ERS), a nonprofit in Watertown, Mass., grapples with that issue regularly. Many school districts “have no reliable way to compare their spending to other districts,” says Dan Turcza, an associate at ERS. There is no standardized reporting system for such information. Comparative per-pupil spending data is available, but it’s not granular enough to show whether a district might be spending too much on maintenance, for example, or spending too little on field trips.
Fixing that problem requires applying uniform labels to hundreds of line items in multiple district budgets. For ERS, doing that much coding by hand poses a huge challenge. That’s where data science—and a competition managed by Driven Data—come into play. “This is a classic machine-learning problem,” says Peter Bull, another cofounder. The contest aims to encourage skilled data scientists to dig into this dilemma, and ERS is offering a $7,500 purse that will be shared by the top three solutions. Driven Data will identify winners on the basis of three criteria: accuracy, speed, and economy. A solution that delivers on those metrics would allow ERS to reduce its prices and greatly extend its impact. (Results will be announced after the contest closes in early January.)
Like their company, the founders of Driven Data span the gap between analyzing data and putting it to use. Bull and a third cofounder, Isaac Slavitt, are recent graduates of a data science master’s program at the Harvard University School of Engineering and Applied Sciences. Lipstein is an MBA student at Harvard Business School. “If we can find interesting questions where data science tools can be used to drive social change, we think that practitioners and students will engage,” Lipstein says.
Driven Data isn’t the first effort to leverage data science for social good, but its approach is different from what others in this young field are doing. DataKind (formerly Data Without Borders) matches data scientists with social impact organizations and helps them conduct quick “data dives.” Bayes Impact places data scientists in yearlong fellowships with civic and nonprofit partners. “There’s an ecosystem coming together,” says Bull. “We see ourselves as part of a continuum.”
What differentiates Driven Data is its commitment to complexity and its competition-based model. The right problem for a Driven Data project is one that’s too complex for a weekend hackathon but not so intricate that it requires an extended research project. “We’re looking for projects with enough complexity that they need to go for two or three months,” Bull says. Online competitions, moreover, can be a powerful tool for developing such projects. “There could be multiple really good ideas for how to approach a problem,” Lipstein says. “We can compare them side by side. It’s not necessarily clear beforehand what the best approach will be.”
Research on open innovation contests supports that view of how beneficial they can be. Not only do such competitions generate an abundance of ideas, but “the value of the best idea generally increases with the variability of the ideas received,” Karim Lakhani and Andrew King write in the fall 2013 issue of MIT Sloan Management Review. (Lakhani, an associate professor of business administration at Harvard Business School, is an advisor to Driven Data.)
Driven Data is now pursuing opportunities in public health, basic scientific research, education, microfinance, and humanitarian aid. Whatever the topic may be, Bull says, the goal is to investigate questions that involve “the most data and the highest potential for social impact.”
Read more stories by Suzie Boss.