War Games (or how the US military learned to love the crowd)

April 26th, 2011 by

microtask_darpa_bigdogQuestion: where do you work if you: a) are a genius, b) are too crazy for NASA and c) really like blowing stuff up? Answer: DARPA.

The US Defense Advanced Research Projects Agency has a $3 billion budget, a remit to “create lasting revolutionary change” and a project list that looks like it was written by Philip K Dick. Current schemes include shape-shifting matter, laser guided bullets, flying submarines and robotic beetles. Since its creation during the cold war, DARPA has been credited with some spectacular innovations including GPS, stealth planes and the first ever internet network. But for every hit, there’s been a (usually embarrassingly wacky) miss. Famous failures include the Vietnam mechanical war elephant and an operation to train telepathic spies. And the latest far-out DARPA scheme? Er, crowdsourcing.

Getting off the ground
Given its cutting-edge reputation, DARPA was surprisingly late to the crowdsourcing party. The agency only launched its first crowd-based experiment in 2009 – an open competition to test whether online social networks could be used to gather precise data. DARPA released 10 red weather balloons at secret locations across the U.S. The first team to pinpoint the coordinates of all of them was offered a $40,000 prize. Using Facebook, Twitter, public radio and good old-fashioned bribery, the winning team tracked down every balloon in an impressive 9 hours.

Obviously impressed by the crowd’s talent for tracking inflatables, in 2011 DARPA launched the snappily titled Experimental Crowd-Derived Combat-Support Vehicle Competition. The competition invited the crowd to submit designs for a “groundbreaking” new military jeep. The contest ran for a month and DARPA received and validated over 150 entries (it’s not a massive number but I guess there are only so many frustrated amateur Humvee designers out there).

Submersive elements
Finally, this April, DARPA waded into the deeper waters of crowdsourced gaming. The Anti-Submarine Warfare Continuous Trail Unmanned Vessel or ACTUV Program is a project to build an automatic submarine-tracking drone. DARPA has asked the crowd to help refine ACTUV’s software by playing a tactics simulation game. In the game players navigate a virtual drone through various scenarios: targeting enemy ships, hiding from commercial traffic, trying to avoid accidentally surfacing off North Korea. After completing missions, players “debrief” their tactics to DARPA. The best crowd strategies will eventually be incorporated into ACTUV’s software.

You might ask why random gamers are testing this simulation instead of say, trained submarine crews. I guess part of the answer is that trained submarine crews are busy crewing submarines. Using the crowd also allows DARPA to collect huge quantities of data very quickly and cheaply. Game scores are also a great way to automatically sort the best and worst underwater tactics. The only limit is the game itself. However imaginative the simulation designers are, real-life sub hunting is bound to throw up a few “untested scenarios”.

The idea of “military crowdsourcing” clearly has some ethical issues. So far, DARPA’s crowd-based experiments have been restricted to tactical and support operations. But what if the next project is helping to design an attack drone or program a flesh-eating robot? The agency seems determined to push the boundaries of crowd collaboration. Time will tell how helpful the crowd is – and whether it will choose to play.


  • http://twitter.com/R3Evolution Ales Nohel

    Thanks for the tip, downloading game now :o )nnNASA also started crowdsourcing with MoonBase Alpha and iPhone game called NASA Lunar Electric Rover SimulatornnDo we have something like DARPA in EU version ?

  • http://twitter.com/cypherinfo g. c.

    Who knows if the DARPA will research and take in examination that: one of the solution to the country defense could be solve issues as: “how to give the third world a path for becoming wealth” and so on.


<<

>>