By Ashley YeagerSoccer penalty kicks, ‘Chicken’ and other games may thwart terrorist attacks, drug smugglers and even freeloaders trying to board trains without tickets.
It’s not so much the intensity and adrenaline of the games that lead to better security, but the logic the players use, says Vincent Conitzer, a professor of computer science and economics at Duke.
This logic is called game theory and now scientists are using it to compute solutions for security issues, Conitzer explained at a July 11 talk with undergraduates completing summer research projects on campus.
During the talk, Conitzer gave a brief overview of game theory using real-world examples, such as penalty kicks in soccer and a set of drivers playing chicken. In the soccer example, he described a “zero-sum game” between the goalie and the kicker, where no matter the outcome, one player wins and the other loses.
But in the case of chicken, in which two cars drive straight at each other until one of the drivers “chickens out” and diverts course, the stakes of each choice are a bit higher. If both drivers stay straight, they crash. It’s no longer a zero-sum game.
When it comes to preventing security problems, there are more angles of attack, smuggler entry points and ways to board a train than the simple left, right or straight of these game examples.To make predictions about what the bad guys will do in the security scenarios, Conitzer is working with Milind Tambe and his group at USC. The team has designed game theory algorithms to set the schedule of security checkpoints and canine rounds at LAX airport, smuggler-scouting in Boston Harbor and even methods for preventing terrorist attacks in Mumbai.
Tambe “treats the problem of Mumbai personally” since that is his home city, Conitzer said, adding that he is only directly involved in this project with the USC group.
While the talk focused mainly on security applications, Conitzer also thinks that some “surprising new applications have yet to emerge” from the work. The new uses won’t necessarily help win a game of chicken or score a penalty kick.
But they could help scientists understand how to better use incentives to designgames with only good outcomes, such as encouraging smart energy use.
Citation: “Computing Game-Theoretic Solutions and Applications to Security.” Conitzer, V. In Proceedings of the 26th National Conference on Artificial Intelligence (AAAI-12), Toronto, ON, Canada, 2011.