AGT in CACM
Tim Roughgarden briefly reviews algorithmic game theory in the latest edition of the CACM . He frames algorithmic game theory as a field addressesing the follows basic questions: To what extent is "incentive-compatible" efficient computation fundamentally less powerful than "classical" efficient computation? When, and in what senses, are game-theoretic equilibria guaranteed to approximately optimize natural objective functions? When can the participants of a game quickly converge to an equilibrium? More modestly, when can a centralized algorithm quickly compute an equilibrium?