What type of team can you organize nationwide to find 10 balloons in 9 hours with today’s technology? That was the challenge posed by DARPA to the nation yesterday. The substantial reward ($40,000) for the first correct entry provided incentive. There were several organized attempts, and plenty of individuals. But only one team widely broadcasted an opportunity with a detailed financial plan for you to participate directly.
Location did not matter. The winning team was headquartered hundreds of miles from the balloons.
Since no one has a social group so large or organized to cover all the possible places (continental U.S. within sight of a roadway) most individuals took an “all or nothing” approach. Individuals concluded that if an organization did not suddenly form to win, the best chance was to troll the blogosphere, twittersphere, websphere etc. for clues.
Others broadcasted the contest with an email invitation to form a private group on the fly. But these seemed to be a way of saying, “you can’t win, so if you see a balloon, tell me where it is.”
Of the teams formed, a group of experienced geocachers, a consolidation of schools, and a team from MIT took different approaches. Geocachers already had a social network and experience finding and broadcasting coordinates. They appeared to take an early lead announcing the number of confirmed balloons they had found. The 10Balloons team organized schools to get their students to find the balloons and report them to a central website with interactive map.
An open website (openredballoon) built an information organizing website and left it open to anyone to report.
The MIT team however, offered specific cash prizes spelled out in detail that were designed to encourage dynamic group formation and reporting. A prize for creating a successful network of searchers and another for each successful finder were described.
DARPA released the balloons on Saturday Dec 5 at 10 AM EST and the balloons remained flying until 4 PM local time. DARPA officials were on hand at each balloon to confirm location information if asked.
During the contest, dynamic reporting included websites, and Twitters that were searchable by hashtags like #redballoon or #10balloons, etc. Backnoise.com was also useful.
As predicted by many tweets during the week leading up to Dec 5, false balloon reports greatly outnumbered the real ones. For example, the four balloons in the southeast were reported with 2 – 4 accurate clusters each within the first 2 hours. Since the balloons were numbered, one could assume that the accurate reports would be clustered around a particular number. But separating them from the noise was impossible.
About 3 hours in, the 10Balloons site became a doodling board, with kids using balloon locations as ‘pixels’ to draw happy faces, circles, and words.
By mid-afternoon, Open balloons was overwhelmed with hits. It appears the site adminstrators had to clear the map on at least one occasion.
Supposed sitings on Twitter in Providence RI and Irvine California looked real for a while, then were discounted.
By afternoon, some people tweeted and posted photos of the DARPA balloons. The balloon in San Francisco’s Union Square (#1) was obvious.
About the same time, news story about the balloon in Tonsler Park, Charlottesville, VA was posted on the web news site readthehook.com complete with coordinates.
The Portland Oregon balloon was reported on Twitter, as well as Santa Barbara’s but without information about exactly where. The rules required reporting the coordinates to within 1 mile, and numbering the balloon correctly.
By 7 PM, nine hours after the contest began, the MIT Team, successfully reported the location of all 10 balloons correctly. DARPA announced the winner and the locations here.
Some possible conclusions:
- A social network can be built quickly if the participants are incented to participate.
- Private social networks form that guard closely held valuable information.
- Any public event will be reported. Some reports will be accurate, and many more will not.
- Twitter is perhaps the best tool today for instant broadcast of breaking events. But breaking news is seldom accurate to any detailed level.
- Unsubstantiated rumors will be circulated. Those with more credibility will be circulated more widely.
- Public sites that invite participation will have a few accurate contributors trying to help, many more who are trying to obfuscate for personal gain (either attention, or increase chaos, or possible reward)
- The most effective social networks offer both an incentive to participate and a charitable contribution.
Feel free to comment on these or other conclusions.