Skip to main content

<em>Science</em>: Cascading Incentives Are the Best Way to Get People Involved

Imagine that you are part of a search-and-rescue team looking for a lost child—or perhaps part of a neighborhood that is looking for a group of outlaws on the lam. What’s the fastest, most effective way to inform other people of the situation and then recruit them for your cause?

It’s a question that many social scientists have been trying to answer. And now, a recent competition held by the Defense Advanced Research Projects Agency (DARPA) is shedding light on the kinds of incentives that can be used to accomplish tasks that require this kind of society-scale level of mobilization.

For the competition, 10 red weather balloons were placed at different locations around the continental United States and competing teams were challenged to locate them all as quickly as possible.

Galen Pickard and colleagues from the Massachusetts Institute of Technology (MIT) team emerged as the winners. In their study that appears in the 28 October issue of the journal Science, the team explains how they were able to use the World Wide Web to find all 10 weather balloons in less than nine hours.

Before the competition even began, the winners were able to recruit nearly 4400 individuals from across the country to help them find the balloons. They accomplished this by not only offering some of the competition’s prize money to anyone who located a balloon, but also by

Video illustration of how referrals between MIT team participants spread on the continental U.S. before and during the DARPA challenge. Each white line represents a referral from an existing participant’s location of the MIT team to the the location of a new team participant. The lines highlighted in yellow represent referrals leading to locations where the balloons were actually launched. | Video courtesy of Wei Pan; © Science/AAAS

offering money to any individual who informed a balloon-finder about the competition in the first place.

“The way we won this challenge was to develop an incentive mechanism,” said Wei Pan, a member of the MIT team. “The idea was: If you found a balloon, you got $2000… but you could still make money if you didn’t find a balloon. The way to do that was to tell a friend to join us. And if your friend found a balloon, they got $2000 and you got $1000 for referring that friend. If your friend’s friend finds a balloon, he or she got $2000, your friend got $1000, and you got $500.”

“This ‘recursive incentive’ mechanism leverages peoples’ social networks,” explained Alex Pentland, a member of the MIT team. “It gets them to actively recruit others to be part of the search. All the other teams only provided individual incentives, so while people who knew about the challenge looked for balloons, they did not actively recruit other people.”

This strategy of using recursive incentives also beat out altruism-based strategies, such as the one attempted by a team from Georgia Tech which offered to donate all of the prize money to the American Red Cross.

“This grand challenge was meant to illustrate that we are at the beginning of creating new types of organizations that are based on peer-to-peer communication and incentives, rather than rigid hierarchy and rules,” said Pentland. “This sort of fluid organization can be better when reacting to disasters and emergencies, and potentially it can be much more agile and efficient than current companies and governments.”


Read the abstract, “Time-Critical Social Mobilization,” by Galen Pickard et al.

Listen to Wei Pan describe the DARPA balloon challenge and the MIT team’s winning strategy.