Original URL: http://www.theregister.co.uk/2009/12/01/darpa_balloon_caper_precursor/

DARPA balloon-hunt compo: Stand by for skulduggery

We may never be told the winners' methods

By Lewis Page

Posted in Science, 1st December 2009 14:18 GMT

The DARPA Network Challenge - a race to find ten large red balloons to be flown at undisclosed locations in the USA this Saturday - is beginning to take shape, with competing teams trying to marshal the legions of crowdsourced operatives necessary for a win. Meanwhile rules and details have been modified in an attempt to prevent some of the more obvious stratagems.

The contest was announced in October to celebrate DARPA's preferred 40th-anniversary-of-the-internet date. It offers a $40k prize to the individual who can first report the locations (accurate to within a mile radius) of the ten 8-foot red balloons which DARPA will fly this Saturday throughout daylight hours. All are to be located in publicly-acessible places within view of roads or highways.

The design of the compo - limited prize money, quite small balloons, all located in areas where they will be seen by ordinary members of the US population - points to crowdsourcing/social networking as the area DARPA is interested in, and indeed the agency's chief Regina Dugan has said:

"The DARPA Network Challenge explores the unprecedented ability of the Internet to bring people together to solve tough problems.”

Various strategies for bringing people together have been proposed. Some organisers, obviously enough, have promised to share the reward with those first to report correct locations - some are even promising to take no cash for themselves, perhaps theorising that the fame of a win will be worth having on its own. Others have stated that the cash will be given to named charities, or even spent "in one way or another... to help promote peace". Others offer payouts for recruiters of spotters. Yet another team promises, if it wins, to spend the entire $40k on making a "giant flying cupcake built entirely out of balloons!"

Meanwhile DARPA has tweaked the contest rules somewhat, introducing various provisions designed to thwart some obvious ploys. Originally a location had only to be accurate to within a second of arc, meaning that each submitted entry would cover a box on the map; this has now been reduced slightly to a mile and changed to a circular radius, making it harder to reliably blanket big areas using lots of submissions. The registration and submission process has also been "designed to prevent automated approaches designed to systematically guess the correct locations", too.

Maybe the nice guys will actually win - but it'll be hard to tell

One can submit ten locations per valid email address. It's pretty certain that some people are right now experimenting to see how many email addresses it's possible to register given the amount of identities and email accounts they can lay hands on - but it presumably won't be enough to try out every possible combination of ten locations. One does note, however, that DARPA merely says that automated entries are "discouraged", not forbidden.

Given that the number of entries a team can practically submit is limited, it becomes worthwhile to feed false information to rival teams. Fake balloons are likely to make an appearance on the day, but DARPA has said that it will announce special authenticating details at the last minute. In any case a team would need to put up much larger numbers of fakes than anyone else to gain a substantial advantage. This would probably prove too expensive to be worthwhile.

Much more likely is infiltration of rival reporting networks to feed fake reports into their systems. A lot of this will probably take place, especially with those groups advertising publicly to gain members. The groups communicating reports by public means such as Twitter can also expect to have their locations hoovered up by rival teams. It also seems qute possible that there are unseen groups which actually control more than one of the publicly-known teams, perhaps in a bid to recruit both mercenary and unselfish reporters.

All in all, there has to be at least a chance that the prize will actually be won by some team which either piggybacks off its' rivals' efforts or actively sabotages or glovepuppets other groups, perhaps combining such methods with large numbers of registrations using many individual identities and/or automation.

A group like this would be unlikely to make its methods public, so there's a sporting chance that details of all the skulduggery will never be known; or not to us, anyway. DARPA would be quite likely to offer an extra, undisclosed sum to a winner for details of methods used, we'd submit, especially if these seemed likely to be of use to the US military.

Alternatively, the prize may actually be scooped by some naive and fluffy group like the charitables or the flying-cupcake mob - though a group of the first and darker sort might also credibly pretend to be of this type.

You have to ask just how much DARPA, or anyway the rest of us, are actually going to learn here. What is for certain is that some interesting - probably largely unseen - shenanigans seem likely to take place this weekend. ®