Agree or Convince

We continue the work of Aumann (Ann. Statist. 4 (1976), 1236-1239), Geanakoplos and Polemarchakis (J. Econ. Theory 28 (1982), 192-200) on common knowledge and consensus, reconsidering the arguments and the findings of both Aumann (1976) and Geanakoplos and Polemarchakis (1982) and offering different insights into the revision process. By revealing set inclusion property of the revision process, we show that the consensus conditions should be redefined. This redefinition enables us to demonstrate that until consensus is reached, in fact neither of the agents make apparent revision and each agent keeps repeating his initial posterior. Therefore the equilibrium posterior should be equal to initial posterior of the agent who does not make any apparent revision through the communication process. Our results show that regardless of the length of the communication process, it is impossible for the agents to agree on a value which is different from the initial posteriors. Finally, we shed light on some crucial points left unclear by Aumann (1976) and Geanakoplos and Polemarchakis (1982).