Game Authority for Robust and Scalable Distributed Selfish Computer Systems
Paper in proceedings, 2007
Game theory analyzes social structures of agents that have freedom
of choice within a moral code. The society allows freedom and
selfishness within the moral code, which social structures enforce,
i.e., legislative, executive, and judicial. Social rules encourage individual
profit from which the entire society gains. Distributed computer
systems can improve their scalability and robustness by using
explicit social structures. We propose using a game authority middleware
for enforcing the moral code on selfish agents.
The power of game theory is in predicting the game outcome for
specific assumptions. The prediction holds as long as the players
cannot tamper with the social structure, or change the rules of the
game, i.e., the prisoner cannot escape from prison in the classical
prisoner dilemma. Therefore, we cannot predict the game outcome
without suitable assumptions on failures and honest selfishness.