Gaming the semi-cooperative

I’m in and around Washington DC for much of this week, and today had the opportunity to give a talk at the RAND’s Center for Gaming in Alexandria. My topic was “gaming the semi-cooperative,” and it focused on the challenge of designing games that are neither purely adversarial (unlike most kinetic, blue-on-red wargames) nor fully cooperative (unlike say Pandemic, or many emergency preparedness exercises). This sort of challenge comes up often in the sorts of games that I develop and use for both educational and policy/analysis purposes: for example, peace processes and operations where most actors may want peace, but they differ greatly in their approach, interests, and vision of the future; humanitarian interagency games, in which players may share a common overarching goal, but also have institutional interests and standard operating procedures that sometimes put them at loggerheads; or even substantially kinetic campaign games characterized by complex and tenuous multinational coalitions of the not-always-willing.

In the presentation I noted that one can try to generate semi-cooperative behaviour through the explicit rewards, payoffs, and game objectives given to the players. This is what might be termed a game-theoretic approach, since it presumes that players will, in rational pursuit of maximum gains and given a particular payoff matrix, adopt the desired semi-cooperative behaviours. And to some extent they will: AFTERSHOCK, for example, deliberately scores players both on their achievement of collective goals (saving lives, represented in the game by “Relief Points”) and separate individual goals (organizational reputation and political or donor support, represented in the game by “Operations Points”), thereby encouraging general cooperation complicated by occasional friction deriving from divergent interests.

The core of my argument, however, is that structuring rewards and explicit objectives is not enough.  Robust evidence from behavioural economics and experimental psychology shows that not all players respond in similar ways to game payoffs: norms, attitudes, and socialization makes a difference. How one frames a game to players has also been shown to have dramatic effect on the proportion of cooperative and non-cooperative actions. From a learning perspective, extrinsic incentives and rewards (scoring points, meeting defined objectives) may be less effective in educational games than intrinsic rewards—the emotional satisfaction—from playing a game well. Role identity and immersion in the game narrative can have powerful effects on game play dynamics. So too do player gaming styles and role assignment importance.

Given all this, I discussed several techniques I have used to manipulate player psychology and narrative engagement so as to foster semi-cooperation:

• The use of limited or manipulated player information to generate friction, rivalry, suspicion, or sense of injustice.
• Time pressures to spur both bonding and friction.
• Manipulation of the physical environment (such as room assignments or game layout) to foster or hamper cooperation.
• Social engineering of participant assignments, using known players in key roles to increase cooperation or tension.
• Recognizing the importance of “fluff and chrome”—that is, the backstory, setting, and supplementary materials of the scenario—in generating a sense of immersion and role identity.

The slides for the talk can be found here (pdf) and here (powerpoint).