Game Theory

By Dawn Lim

April 1, 2012

Last Halloween, while his friends were out reveling, Otavio Good stayed home to design a computer program to reconstruct shredded documents. The military’s research arm had issued a call for people to decipher messages that had been torn to bits, much the way someone might destroy evidence of a paper trail. The Defense Advanced Research Projects Agency’s goal: to develop new intelligence-gathering methods. Good, a 37-year-old designer of mobile software applications, had slightly different goals: “fun and worldwide bragging rights.”

Teaming up with eight Silicon Valley friends and programmers, Good refined a script that processed the ink and contours at the edges of the shredded pieces, matching them up. Deploying a multicore processor to get ample computing power, the group began running and tinkering with the script after work and through weekends. “Is your girlfriend OK with your spending all your time on this?” Good asked a teammate. The response: “We’re still in first place, so that helps.”
 
While his team labored to solve the challenge, DARPA used a score card to track the progress of 9,000 teams.
 
The eureka moment came when Good and his teammates created a computational method to map out the paper’s watermarks and stitch together its pieces. Six hundred hours and 33 days later, they pieced together five puzzles and 10,000 shreds, winning $50,000. DARPA flew the team, dubbed All Your Shreds Are Belong to U.S.—a play on a badly translated online video game that went viral in 2000—to Washington. There, military officials signaled interest in exploring ways to adopt what the team had created. 
 
Critical Mass 
 
The shredder challenge is one of dozens of experiments in crowdsourcing that federal agencies are performing to tap the expertise of people they wouldn’t otherwise reach. Military boardrooms and government laboratories aren’t always the most conducive spaces for flashes of insight and creative thought. Nor do they attract Silicon Valley types. But by mining the crowd for answers agencies can’t find on their own—with games that reward ingenuity and play—they are accessing a wealth of ingenuity beyond the civil servants, military personnel and contractors who comprise the federal workforce. Freed from the constraints of reality, gamers are able to conjure ideas that an expert in a cubicle might never think of. If these games and puzzles feed bright ideas to government leaders, they could upend the perceptions many people hold about computer games, from black holes that suck resources from society to tools with real-world impact. 
 
Foldit, a computer game partially funded by DARPA and the National Science Foundation, showed how good game design can channel the passion for playing into answering scientific questions. Hundreds of players jiggle protein molecules on Foldit each day, scored by how efficiently their creations assume their structures. “Foldit turned novices into experts,” says Zoran Popovic, director of the University of Washington’s Center for Game Science, who led its development. When a gamer with the handle “Mimi” made the last move that deciphered a protein crucial to replication of simian AIDs that had baffled scientists for decades, her mouse click sealed Foldit’s iconic status. Still, the team she collaborated with resisted being identified in research papers; they wanted to keep folding proteins in obscurity. Fame never crossed their minds when they started playing. “I thought it would be of more benefit than wasting time on a game that was amusing but not accomplishing anything,” Mimi, a part-time school laboratory technician, said on Skype from Manchester, England. 
 
The self-effacement of the winning team masked a subversive victory. Citizen “crowdsourcing and scientific discovery really challenge notions of expertise that are fine for some, but uncomfortable for others,” says Constance Steinkuehler, a senior policy analyst in the White House Office of Science and Technology Policy. Steinkuehler, a game researcher on an 18-month stint from the University of Wisconsin-Madison, aims to steer the Obama administration to get serious about games. 
 
If government leaders can overcome  biases against games and crowdsourcing, there’s untapped brainpower at stake: In the United States alone, 72 percent of households play computer or video games, according to a 2011 report by the Entertainment Software Association. Government could boost a growing game industry that’s already racking up more than $25 billion in annual sales, according to the ESA report “Essential Facts About the Computer and Video Game Industry.” In the United States there are an estimated 145 million active gamers and 215 million hours a day are spent on game play, according to market research group Newzoo. The question remains, however, is the government ready to take advantage of these trends? 
 
Military Crowdsourcing
 
After Foldit’s success, Popovic demonstrated to DARPA, the military’s venture capital arm, that it was possible to create a game to test for software defects. The concept: to map software code visually as a network of pipes on a screen. Players then would send as many balls as possible through the pipes, resizing and cutting the tubes to prevent them from getting stuck. Any spot where a ball got stuck located a specific point in the code that had potential security glitches. 
 
That proof-of-concept prompted DARPA to invest $32 million in developing “fun to play” games for laptops and smartphones that help debug software code. The three-year experiment, called Crowdsourced Formal Verification, aims to help the Pentagon cut costs while it grapples with a shortage of security specialists to test weapons systems software. “No one wants to look at code,” says Popovic, “but if you cast the problem visually, more would tackle it, not just a few experts.” A crucial part of game design involves ensuring the safety and anonymity of government agencies that submit code for verification. 
 
The Navy also hasexperimented with crowd­sourcing techniques through a game that is inelegantly known as MMOWGLI, or Massive Multiplayer Online War Game Leveraging the Internet. It was created by the Office of Naval Research and partner organizations to leverage the crowd to develop an effective response to Somali piracy. Players generate cards with suggestions of less than 140 characters on ways to fight piracy that other gamers help score. Two programmers and multiple design reviewers spent 18 months developing the game, says Don Brutzman, a retired submarine officer and associate professor of applied science who led its development at the Naval Postgraduate School. Within 10 weeks of its launch in May 2011, 16,000 people signed up and 832 played. The game garnered futuristic suggestions such as, “use radio frequency identification plants in sacrifice decoys to track booty and pirate traffic, then surprise leaders with proof,” as well as feedback from users on how to improve any less-than-user-friendly interfaces. 
 
“The jury is still out on the most relevant solution the game offered,” admits Lawrence Schuette, director of ONR’s Office of Innovation. He declined to provide funding details. “The real relevancy is that a broader community exists out there to source ideas from.” Energy MMOWGLI, expected to be released this month, will focus on how the Navy can meet energy demands. Developers are working on ways to visualize and simulate some of the plans, paving the way for MMOWGLI to transition into a more interactive platform. 
 
“The traditional Navy analyst community did question if you could do anything out of an online game,” says Schuette. “We took science fiction to science fact.” 
 
The Games Guru
 
The Navy’s efforts are laudable considering how hard it is for agencies to justify projects that may have little return on investment. “[Unlike] the typical indie game developer who will sleep on friends’ couches, if necessary, to put out a beautiful, compelling game,” says Steinkuehler, “game development in the government is done by people on top of a whole lot of complicated stuff on a workday.” Steinkuehler has the task of invigorating the nascent government game development community so that when she leaves her post in mid-2013, she will have created lasting momentum. 
 
“If the market is un­able to get the [financial] return it needs to make bleeding-edge games, such as discovery and crowdsourcing games, one of the roles of government is to seed these spaces,” she says.
 
Steinkuehler convened a meeting in November 2011 for agencies to troubleshoot issues around deploying games. Only 40 people were expected, but 70 showed up. It was community building for people who had been working in isolation for years. As of March, following two more meetings, the interagency working group had grown to 156 people from 33 agencies and four White House offices, with various levels of expertise. And they’re not coming for the refreshments: No snacks are provided and participants bring their own coffee. “We’re still writing up the notes and trying to see if we all heard the same thing at the meetings,” says Daniel Laughlin, a NASA research scientist and project manager.
 
Steinkuehler’s presence is a stamp of approval for federal scientists who have been trying to secure more recognition for this kind of work. For some, it also has been worrisome. “People are afraid they’ll miss the train and their agency will be the laughingstock of the government game community,” says Laughlin. Many are responding with caution. “I would imagine continuation of the group depends on who gets elected [to the White House] next,” says Eric Hackathorn, a data visualizations and games program manager at the National Oceanic and Atmospheric Administration.  
 
Learning to Teleport
 
Both Laughlin and Hackathorn know that game development in government is a process that requires working around forces beyond their control. 
 
Laughlin, who has been pressing NASA to support games that promote science and learning since 2004—by churning out reports and PowerPoint presentations—is realistic about funding constraints. “The chances of NASA providing the money that World of Warcraft took to develop are as much as the chances of me learning how to teleport,” he says of the enormously popular game. 
 
To work around this, NASA found a commercial team that would raise capital and devise a business model for a space-themed, multiplayer game. Three video game companies, WisdomTools, Virtual Heroes and Project Whitecard, were named as partners in December 2008. NASA forked over $1.65 million for its experts to work with developers to incorporate agency data sets and educational content into the game. The three companies had to form a legal entity first because NASA demanded a structure that would act in concert before it would sign a deal. The partnership to create Astronaut: Moon, Mars and Beyond was finally inked in August 2010. The first module of the game is expected to launch next year. 
 
The success of an earlier NASA-funded game powered Laughlin through the bureaucratic hurdles. Moonbase Alpha, a prototype to show that NASA data could be overlaid onto a multiplayer game, was launched in July 2010, hitting 100,000 downloads in its first month. It still gets about 25,000 downloads a month. The proof-of-concept also generated ideas about how crowdsourcing could be woven into games. The day after Moonbase Alpha was released, players surprised programmers by flipping the lunar exploration rover, which had been modeled after actual vehicles. This highlighted the possibilities of using games for space equipment testing and science projects. At NASA, volunteers have been tapped to find interstellar dust grains captured in aerogel collectors on a spacecraft and mark craters on footage taken on space missions to build a map of Mars.
 
Hackathorn’s efforts to promote games at NOAA show the delicate negotiations that sometimes are required for support. After years of pitching a game to tap the crowd for improving weather prediction models, he couldn’t get enough funding and had to go back to the drawing board. “I’ve had a few people tell me they were worried such an idea could take jobs away from meteorologists,” he says. 
 
To help mollify skeptics, Hackathorn focused on creating interactive spaces for scientific learning. An experiment establishing an island in virtual world Second Life where users learn about meteorology proved less than favorable. “Even though you own the content you create, it is nearly impossible to remove it from their system and use it elsewhere,” he says. A spokes-
person for Linden Lab, the company that created Second Life, says, “federal organizations have used Second Life as other organizations have in the past, generally without Linden Lab making changes to the platform specifically for them.” When the hype around Second Life fizzled and traffic dropped, Hackathorn’s team focused their attention on other game engines where vendor content wouldn’t be in a format that could be read only on the company’s platform. “If a game engine goes belly-up, I want to still have the content I’ve created,” Hackathorn says. 
 
Fragile Earth Studios, a NOAA software development studio he leads, eventually got tens of thousands of dollars in funding to create TerraViz, a project that visualizes real-time weather, earthquake and ozone data on a virtual globe using the development tool called Unity 3D. The hope is TerraViz will start to have gamelike elements, rewarding the efforts of people who dive in and analyze that data. Hackathorn’s team is working on ways for users to annotate the globe, and creating a virtual space where users can take on avatars. “We will tackle crowdsourcing when it’s more politically favorable to look at it,” he says. 
 
Reengineering the Expert
 
An Army project to study the irrationality of group conduct could offer a counter-point to the argument that crowds are infinitely wise, helping to frame crowdsourcing in a less threatening way. The project, Virtual Laboratory of Aggregate Behavior, will create a platform “to facilitate the execution of random control trial experiments that require the real-time, simultaneous participation of hundreds or thousands or more participants,” notes the call for research proposals. By studying the churn of the crowd, scientists could better understand how to control protests, neutralize the spread of extremism through online networks and avert crises. The Army hopes military planners and companies could use this to answer questions related to defense, marketing, health care and urban planning.
 
Technical challenges lie ahead. It will be difficult to stretch a grid that can host so many people. It will require skill designing an environment where multiple experiments can take place, while allowing each experiment to be specific enough to be useful. To have the best game designers, the government funding process has to be more transparent. Developers and bureaucrats have to learn to speak the same language. Vaporware, the term used when software fails to live up to its hype and is never delivered, remains a concern for agencies wading into emerging game technologies. 
 
If it wants to embrace a crowdsourced future, the government will have to learn to seek out help and be challenged. This doesn’t mean expertise is dead, says Hackathorn. “It simply will be a reengineering of what an expert is.” 
 
Dawn Lim, a journalist in New York, is a former intern for Government Executive.

By Dawn Lim

April 1, 2012

http://www.govexec.com/magazine/features/2012/04/game-theory/41624/