On January 27, Microsoft Research released its second Facebook game "Doubloon Dash: A Quest for Understanding," aimed at helping to understand users' behavior in social-networking situations.
The goal of the Research Games project is to "test the behavior of real people in game-theoretic interactions, particularly those that occur in social networks."
"This game models situations like job interviews, patent races, or competitions,' explains (Thore) Graepel, a principal researcher with the Machine Learning and Perception group at the U.K. lab. 'In all of these cases, people are effectively playing an all-pay auction where everyone loses their bid but only one player gets the reward.'"
"The critical question then is: How much should you bet? If you bet more, your chances of winning are higher, but you are risking more money for a smaller reward. If you bet less, your chances of winning are smaller, but you risk less money and if you win, you win more. And then, there is your opponent. What will he or she do? Can you outsmart your opponent? The game, though, is less about beating your opponent and more about making the best of the situation--to earn as many doubloons as you can," notes Inside Microsoft Research blog.
The researchers behind it are hoping to capitalize on Facebook's viral-marketing mechanisms so as to be able to observe a large study base and to analyze better the "playing field" of the social graph. "The artificial 'lab setting' of traditional experiments in behavioural game theory is replaced by the natural social habitat of the subjects, their circle of friends," the Microsoft Research UK page notes.
The nature of the new game presents players with an intriguing set of choices.
"With Doubloon Dash," Graepel says, "we are hoping to learn how people behave in this strange type of all-pay auction. Are they careful and bid little, or are they reckless and bid a lot? How are they adapting their behavior to their opponents and over time?"