演講者簡介 : Professor Manuel Mueller-Frank received his Ph.D. in Economics from Northwestern University in 2010. He is currently a Professor at the University of Navarra. Her research interests are Game Theory, Economics of Information, and Blockchain and Cryptocurrencies.
演講摘要 : The game theoretic analysis of blockchain oracle mechanism so far has explicitly (or implicitly) assumed that reporting agents commonly know the realized state of the world. We investigate the robustness of oracle mechanisms to perturbations to common knowledge. A Bayes Nash equilibrium is continuous if for any sequence of types converging to common knowledge, the equilibrium strategies converge to the equilibrium strategies at complete information. An oracle mechanism satisfies strong continuous implementation if all of its continuous BNE implement the true state. We show that any oracle mechanism that does not provide payments to reporting agents in an oracle native coin, fails strong continuous implementation. We further show that any (simultaneous-report) oracle mechanism that satisfies strong continuous implementation is either dictatorial or vulnerable to single deviations, where a single agent can (probabilistically) prevent the implementation of the true state. To address this weakness, we propose a sequential direct voting mechanism. We show that this sequential mechanism satisfies strong sequential continuous implementation and is not vulnerable to single deviations.
