Analysis of Opinion Aggregation Method Using Metacognition on Social Networks

Abstract

We analyze an opinion aggregation method that uses metacognition on social networks. While the majority vote is widely applied, it does not always result in a correct decision when the majority has incorrect opinions. As one solution, a new voting method called surprisingly popular (SP) recently has been proposed by utilizing metacognition. Some existing works have evaluated the performance of the SP by using a crowdsourcing service. However, there exist no analyses of how well the SP performs when people on social networks decide the social decision by using the SP. We show that the performance of the SP depends on (1) the correctness of the metacognition and (2) the structure of social networks in computational simulations.