Social media skills put to test in 'diamond heist'

An international challenge organised by the US State Department and won by a team from the Masdar Institute was created to show the power of 'crowdsourcing'.

One of the world's most valuable diamonds has been stolen by a gang of five robbers.

After making their escape, they split up and scattered to five cities around the world. A team of international detectives has just 12 hours to track them down, using only the power of social media.

And that's the giveaway to this getaway. The crime is not real; instead, it is the focus of an international challenge, organised by the US State Department and intended to show the power of "crowdsourcing" - a way of outsourcing tasks to a distributed group of people. And it was won by a team from Abu Dhabi's Masdar Institute.

The chase began on March 31, at 10am UAE time - 8am in Bratislava and Stockholm - with the release on the challenge website of two suspects' mugshots.

It wasn't much to go on, recalls Dr Iyad Rahwan, the leader of Team Crowdscanner and an associate professor of computing and information science at the institute. Nor did they have much idea, initially, who they were up against.

"We knew we had competition from other teams but we didn't exactly know how many," he recalls. "But from checking Twitter and other social media I got the sense that there were at least four other teams."

Also on their team was the University of California in San Diego's Dr Manuel Cebrian, a veteran of the Massachusetts Institute of Technology (MIT) team that won a similar competition in 2009.

Then, competitors had to track down red balloons across the US, using a system of financial incentives for people who gave a useful tip.

"We took that idea and developed it further," said Dr Rahwan. "It's all very well having people on social media tweeting messages, but to really focus their attention and to get them to actually work for a specific goal, financial incentives are the best way."

Borrowing a technique from network marketers, those signing up to the challenge were given $1 for each person they recruited.

The second person recruited in the chain was offered $100 for snapping and uploading a photo of a suspect. If the original recruiter managed to spot a suspect and upload a picture, they were given $500.

"In this way, people stayed motivated to be on the lookout for the suspects.

"At the end of the day, the only thing they were giving up was their time and they had the potential to make a bit of money by being a bit more observant during the day."

The first breakthrough came at 4.18pm (UAE), when Crowdscanner were sent three pictures of the Bratislava suspect.

That was a surprise; their hopes for Bratislava had been low because English is not widely spoken there, and none of the team spoke Slovak.

"We quickly asked the volunteers for contact details, and called him to make sure there was no sabotage," said Dr Rahwan.

The pictures were submitted to the organisers and posted on the team's Twitter and Facebook feeds, and their own website.

And then things went quiet. Two suspects were still on the loose in London and Stockholm. As the hours passed, the team sat helplessly in their offices in Abu Dhabi, California and Southampton, unable to do anything beyond checking up on their rivals to see if they were doing any better.

Then at 11.23pm (UAE) a volunteer spotted the second member of the gang at Grand Central Station in New York.

"This felt really good - it proved the one in Bratislava wasn't a fluke.

"The great thing was that we still hadn't had any form of announcement from any of the other teams."

But a couple of hours later, the rival TagTeam - a US-based team - said they, too, had found the New York suspect. At 3am (UAE), the team received a surprise email from Dr David Alan Grier, a professor of computer science at George Washington University in Washington, DC, telling them he had snapped another of the targets.

That, said Dr Rahwan, was "pretty astonishing" for someone so senior in his field.

The competition was hotting up. Once Dr Grier tweeted that he had made the find, TagTeam asked him for the picture. Fortunately for Crowdscanner, he ignored them.

That was not the only attempt at skulduggery - the team had to fend off apparent bids to lead them astray, with red herrings and even sabotage.

At one point a user sent in four emails in rapid succession, each contain a computer programme that appeared designed to tamper with critical parts of a PC's operating system.

"This was assessed as, at best, an attempt to annoy us, or at worst, an extremely amateur attempt to make our systems vulnerable to an attack," Dr Rahwan said. "Their affiliation is unknown, but malicious intent could not be ruled out."

TagTeam had a similar incentive system to Crowdscanner, with $400 to the finder, $100 for referrers, and the remaining $500 per suspect going to charity.

Where TagTeam went wrong, says Dr Rahwan, was in focusing too much on Twitter and then spamming individuals as well as social, government and private organisations in the target cities. Such pleas were often ignored.

And with three suspects found, Crowdscanner were well ahead. Two of their rivals had found one each - but in each case Crowdscanner had beaten them to the punch. On April 4, the competition organisers declared - on Twitter, of course - that Crowdscanner were the winners.

The victory under their belts, the team have turned their attention to what they can learn from their successes and failures.

"We were thrilled to have found three out of five suspects, and could not quite believe it had been done," said Dr Victor Naroditskiy, a Crowdscanner member from the UK's University of Southampton. "With the right set of tools, unthinkable information-gathering tasks can be accomplished in a timely manner."

Prof Nick Jennings, also at Southampton, put the win down to the team's website. Built before the competition started, it contained links and information about Crowdscanner.

"This contributed to establishing the credibility of our team, which is extremely important for the success of online projects," he said.

And the organisers are well pleased with the outcome. It will, said Steve Miller, a co-organiser of the challenge, "provide plenty of ammunition for both sides of the social media debate.

"No one was able to find all five individuals. There may be limits to what social media can help people accomplish in time-critical situations, or internationally."