Bursting a Few Balloons Regarding the Famous DARPA Red Balloon Challenge

It is approximately 50 years since the first appearance of the ARPAnet, the predecessor to the now ubiquitous Internet. It is timely to reflect back on the DARPA Network Challenge (also known as the Red Balloon Challenge) in 2009 in honor of the 40^th anniversary of the first remote log-in to the ARPANet on October 29, 1969. The results of that competition were reported in Communications² but some important details have not previously been reported.

Teams had to find 10 red weather balloons deployed at undisclosed locations across the continental U.S. The first team to correctly identify the locations of all 10 would win a $40,000 prize. It was intended to explore how the Internet and social networking could be used to solve a distributed, time-critical, geo-location problem.

A team from the Massachusetts Institute of Technology (MIT) won in the surprisingly short time of less than nine hours. Much was learned regarding the value of social media, crowdsourcing, and incentive schemes. At the time, social media was rather new and these were valuable lessons learned.

But, there were several aspects of the event that did not get much, if any, attention. Much had been highlighted about the importance of the hierarchical incentive scheme used by the MIT team¹ and how that helped them assemble such a large and geographically diverse team—leading to their success. That is, each member who joined the MIT team was encouraged to recruit additional members who would then recruit further additional members, and so forth. The reward for finding a red balloon not only went to the person who found it, but also to the chain of people who recruited the finder.^a Thus, creating a strong incentive to grow as large a network as possible. In fact, it allowed even people living outside the U.S. to participate. Although they would not be able to personally locate a balloon, they could sign up friends and family in the U.S. to do the searching and gain some reward. In essence, the event was a form of a global pyramid scheme.

That is all quite exciting, but the reality may be somewhat different. In talking with people who were part of the DARPA group running the event and those who were part of the MIT team, it seems that instead of "People finding the balloons," it was much about the balloons finding the people.

Back to Top

Balloons Finding the People

It was reported in Tang et al.^2,b that "DARPA selected readily accessible public sites where the balloons would be visible from nearby roads, each staffed by a DARPA agent who would issue a certificate validating each balloon location."

The consequences of this was not reported. It should not be surprising that often someone, totally unaware of the DARPA challenge, would be walking in these public places and see a big red balloon tethered to the ground. They might ask the attendant (the DARPA agent): "What was going on?" It is not clear how much would be explained, but the attendant would hand out a certificate with information including the precise coordinates of the balloon—note that not everyone carried around GPS systems in 2009 (the accompanying image depicts a certificate).

That person could then use the Internet to learn about the challenge, typically search around and find that the MIT team was still recruiting and sharing the rewards.^c They could then rapidly join the MIT team and report their find to DARPA—and become part of the winners, since finding just one balloon provided no reward, you had to be part of a team that found all 10.

This phenomenon was verified by some of the members of the MIT team who noticed some of the winners did not join their team until after the competition had started and almost immediately before they reported finding the balloon. Of course, joining after the competition had started might not be surprising if it had run for several days as planned, but it started and ended within seven hours.^d Even more curious was the close timing of joining and reporting the balloon.

Ideally, it would have been good if the balloon finders had been individually interviewed to determine how many joined the team first and then sought the balloons versus those who found the balloon and then sought a team to join. Unfortunately, such information was not reported—that would be a good idea for any future such research.

This distinction is important because some proposed uses for such crowdsourcing and the MIT incentive scheme are for things like locating runaway children or criminals. Although forming a network to help search for the child might succeed, you probably cannot count on the child or criminal constantly shouting: "Hello, I am here!" the way the balloons and their DARPA attendants essentially did, which likely led to the rapid speed of locating.

The incentive scheme had a number of other interesting effects, some of which have been reported in the past. It has also inspired similar competitions. To illustrate that, here is one not previously reported. In 2011, a competition was launched to find five missing Langley Knights in England.^e It differed in that not only were there knights to be found in various physical locations in England but also knights that had been placed on the Google Earth Map—so that people from around the world could participate both as recruiters and as searchers.

These challenges have demonstrated a number of features that can truly help to exploit the power of the Internet and social media.

In this case, the winners were interviewed, which revealed some interesting facts. A husband admitted that he had formed a team to seek knights. But rather than report the knight that he found immediately, he arranged for his wife to join his team and then she reported the find. As a result, both he and his wife got a reward, rather than just one of them.^f

In conclusion, these challenges have demonstrated a number of features that can truly help to exploit the power of the Internet and social media. But, it is also important to be prepared for unexpected behavior and results and include methods, such as interviews, to discover them.