The Defense Advanced Research Projects Agency’s “Shredder Challenge” could be called the world’s ultimate jigsaw puzzle contest. But it was serious business for the nearly 9,000 teams of problem solvers from all over the world who entered the competition after DARPA officials launched it in late October.

In the Shredder Challenge competitors attempted to reconstruct machine-shredded documents in increasingly difficult stages to claim DARPA’s $50,000 prize. The challenge comprised five separate puzzles in which the number of documents, the content of the documents and the technique employed to shred the documents varied to make the challenge progressively more difficult.

A small team from San Francisco, calling themselves “All Your Shreds Are Belong to U.S.,” was named the winner on Dec. 2, having correctly reconstructed each of the five documents and solved their associated puzzles. The team devoted nearly 600 man hours developing computer-vision algorithms that suggested fragment pairings, which were then verified manually by team “assemblers.”

The competition drew a variety of teams, ranging from hardcore jigsaw-puzzle solvers to computer scientists and academicians. The results proved highly promising, officials said.

“Lots of experts were skeptical that a solution could be produced at all, let alone within the short time frame,” said Dan Kaufman, director of DARPA’s Information Innovation Office. The winning team solved all five puzzles in a about a month.

But DARPA saw “almost immediate returns on the first puzzle, within a few days,” Kaufman said. “Some real clever jigsaw puzzle fans, I think, are the ones who stepped up.”

The challenge had a practical objective. U.S. military troops often seize the remnants of destroyed documents in war zones but restoring them is problematical and complex. The goal of the Shredder Challenge was to identify and evaluate potential techniques for putting shredded documents back together and making their content intelligible.

Kaufman said the idea was to explore whether shredded documents could be restored in a “military relevant” environment, “meaning can you do it fast enough and cheap enough and within a mission timeframe.”

DARPA’s ultimate objective is to develop an agile, fully automated system for reconstructing destroyed document quickly in the field, said Norm Whitaker, deputy director of the Information Innovation Office. “If we could get that, that would be really great,” he said.

Perhaps surprisingly, the non-technological approach-painstakingly putting the pieces back together by hand-has worked for a number of contestants. “Sometimes, manual is actually faster,” Whitaker said. “In fact the first [puzzles] were solved that way.”

Overall, the most effective approaches were not purely computer-based or human-based but “a combination blended with some clever detective work,” Kaufman said.

To be sure, the challenge to some extent illuminated the differences between human intelligence and computer capabilities. For example, Kaufman noted, humans are “particularly strong” at contextual or intuitive understanding when observing the “entire design space” of a shredded document.

But as the puzzles got tougher, especially for the manual jigsaw-puzzle solvers, and comprised thousands of fragments, contestants applied more automated techniques. Three of the top five teams on the challenge’s leader board used automated or semi-automated approaches that included human involvement, Whitaker said.

Several top contestants deployed crowd-sourcing techniques, where they enlisted hundreds of other problem solvers in their effort, in effect compounding DARPA’s own crowd-sourcing approach to the shredder problem.

The agency has a long history of staging crowd-sourcing-type events, having been granted congressional authority in 1999 to put on problem-solving competitions, Whitaker said. Noteworthy competitions have included the DARPA Grand Challenge in 2004, an event designed to determine the capabilities of driverless cars, and the Network Challenge in 2009, a contest for exploring the roles the Internet and social networking play in the real-time communications, wide-area collaborations and practical actions required to solve broad-scope, time-critical problems.

DARPA experience has spawned a broader wave of similar crowd-sourcing efforts across government, supported by a program known as Challenge.gov, a turnkey system that lets agencies administer their own innovation contests and challenges. Challenge.gov has helped 36 federal agencies launch 130 challenges over the past year.

“DARPA in general gets a lot of things out of these kinds of challenges,” said Whitaker. “One thing we’ve learned from crowd-sourcing is that the people out there are incredibly smart and capable.”

Officials at DARPA also have found through crowd-sourcing that people in the U.S. are eager to participate in projects that support the warfighter. “We have lots of people who watch what we do and sign up strictly for that purpose because that want to be involved in things that advance technology that helps protect [soldiers] on the battlefield,” Whitaker said.

However, Kaufman cautioned, crowd-sourcing is not a panacea. “It won’t solve all problems for all people but we are continually impressed, pleased and interested in the breadth of problems that can be formulated in a way that’s accessible to the crowd,” he said.

Whitaker described the Shredder Challenge as “a first crack” at the problem of reconstructing document fragments. “We’re a [research and development] organization,” he said. “What we do doesn’t always directly relate to a product and it’s not always a direct analysis of a capability. So this first look is trying to understand what’s involved with [an] unshredding task.”