#1157 2009 · MIT Media Lab (DARPA Network Challenge team) · Research / social networking
To find 10 balloons across the US, MIT paid strangers to recruit strangers instead
问题
DARPA hid 10 balloons across the entire continental US; no team could physically search that much ground in a day
背景
In December 2009, DARPA launched the Network Challenge, tethering ten 8-foot red weather balloons at fixed, visible public locations scattered across the continental United States and offering $40,000 to whichever team could report all ten locations first. No group, however large, could plausibly dispatch searchers to cover a landmass that size on short notice; DARPA's actual research question was whether internet social networks could mobilize wide-area, time-critical collective action fast enough to solve it anyway.
A team from MIT's Media Lab entered with a strategy aimed not at searching but at recruiting searchers through cascading incentives. Rather than offering a flat reward only to whoever spotted a balloon, they designed a payout that split $2,000 per balloon down the referral chain: the spotter got $2,000, whoever had invited that spotter into the network got $1,000, whoever had invited the inviter got $500, and so on back up the chain, so anyone forwarding the challenge to friends stood to profit even if they never found a balloon themselves.
换别人会怎么做
The available approach was to organize as large a search operation as the team could staff or crowdsource directly, canvassing likely locations with volunteers coordinated centrally, an approach every other entrant used and that scales only as fast as a team can recruit and deploy searchers itself.
他们看到了什么
MIT saw that paying only the finder rewards search effort, but paying the whole recruitment chain rewards network growth itself, and the real bottleneck was spreading awareness fast, not finding balloons.
那一手
The MIT team offered a recursive referral reward — a spotter's $2,000 split down the chain of people who had recruited them into the effort — which incentivized participants to recruit as many others as possible into a rapidly branching network, since every successful referral anywhere down the chain paid the recruiter something.
为什么管用
Because a share of the balloon reward flowed back through the entire referral chain rather than only to the finder, the incentive to forward the challenge remained strong even for people with no realistic chance of spotting a balloon themselves, which meant the network could branch and cover ground exponentially rather than linearly. This let the search area scale with how fast the recruitment chain could grow rather than with how many searchers the organizing team itself could deploy, which is why a research lab beat thousands of teams running conventional coordinated search.
值了多少
MIT's team found all 10 balloons in under nine hours, beating over 4,000 rival teams that relied on physical search parties.
什么时候会失灵
The mechanism depends on a reward pool large enough to remain meaningful once split across a growing chain of intermediaries, and on a task that genuinely benefits from wide, fast network propagation rather than depth of local search effort; if the prize is too small to motivate forwarding once divided, or the target requires sustained expert search rather than a lucky sighting, a recursive referral incentive adds coordination overhead without a matching payoff.
后来呢
The Red Balloon Challenge became a foundational case study in network science and mechanism design for crowd mobilization, cited in subsequent DARPA-style challenges and academic research on how recursive incentive structures can convert a slow, geographically bound search into a fast, self-propagating one.
资料来源
- [1]MIT wins DARPA's Great Red Balloon HuntNetwork World, 2009networkworld.com
- [2]MIT-based team wins DARPA's Red Balloon Challenge, demonstrates power of social networks (and cold hard cash)Engadget, 2009engadget.com