This article is more than 1 year old
MIT win DARPA's flashcrowd balloon-hunt compo
Asymptotic free lunches mobilize spotter horde
An MIT-organised crowdsourced network of spotters was the first to locate ten large red balloons placed around the USA on Saturday, winning a $40k prize from DARPA. Most of the cash will be distributed among the network's members according to a simple formula whose design lay behind the MIT team's success.
DARPA announced the balloon-finding competition to commemorate its chosen 40th birthday of the internet, and to investigate the ways that online social networks can be used to solve problems. The Agency announced victory for the MIT team on Saturday, saying it had taken less than 9 hours to correctly locate all ten balloons.
“The Challenge has captured the imagination of people around the world, is rich with scientific intrigue, and, we hope, is part of a growing 'renaissance of wonder' throughout the nation," said DARPA chief Regina Dugan. “DARPA salutes the MIT team.”
The MIT team used an elegant and fairly simple means of recruiting its large nationwide network of spotters. Every person registering with the team was given a unique link, which they could then propagate via email, webpage, social network or any other means they saw fit. New people could then click on this link and register with the MIT team, and in turn be given their own unique links with which to recruit others. The MIT database would thus know who had recruited whom.
On Saturday, a spotter managing to be first to get a correct location and balloon number to MIT came in line for a $2000 payout from the team. The rest of the $4k prize money per balloon was assigned asymptotically* to the chain of people who found the spotter: $1k to the person who recruited the spotter him/herself, $500 to the person who recruited the recruiter, $250 to the person who recruited them and so on until everyone in the chain was paid off. The money remaining from each $4k, probably quite a small amount, will be donated to charity.
The cunning of this method is that it makes it worth someone's while to join the MIT team, even though there is only a vanishingly small chance of actually finding a balloon oneself. For a few minutes' work doing a signup and dropping a link onto one's website** - perhaps followed up with an email/IM to your whole address book, for the old-school inclined - you come in line for potentially fifty, a hundred, maybe even a thousand dollars without so much as looking out of the window. MIT will no doubt have benefited, compared to other teams trying similar payout-splitting methods, from having an established reputation (both for probity and technical competence), a high profile among the sort of people likely to be interested and to have large online followings, and no profit motive.
As predicted, there don't seem to have been many fake balloons, and such as there were were most probably bowled out by DARPA's release of pictures just before the event showing what the real ones looked like.
The ten locations, shown here, were mostly in coastal states - none in the landlocked heart of the USA, though there was one in Tennessee. That may have been a wise call by DARPA, seeking to investigate the spread of information among trendy digital America. Had the Pentagon boffins put up some balloons in Kansas, Nebraska or Wyoming, for instance, the victory might not have been quite as swift. ®
Bootnotes
*That is, all the money would be handed out only in the case of an infinite number of people in the chain.
**An ordinary one, or one of the increasingly easy-to-operate automated websites nowadays so famous as blogs, social networks, microblogs etc.