Михаил Ануфриев (Economics Department, UTS Business School University of Technology Sydney) представит статью “(Re-)inventing the Traffic-Light: Experimental Evidence on the Design of Recommendation Devices” (авторы: Mikhail Anufriev, John Duffy, Valentyn Panchenko and Benjamin Young).
The payoffs that players can earn in non-cooperative games may be improved if the players have access to an external recommendation device such as a traffic light that allows their strategies to be correlated. Indeed, the use of recommendation devices is often associated with the concept of correlated equilibrium, an important generalization of Nash equilibrium. In this paper, we report on an experiment in which subjects are incentivized to design recommendation devices for the play of several different 2 player non-cooperative games. Subjects design these devices by colouring balls in an urn in a manner that allows the recommendations made to the two players to be correlated. Recommendations are then repeatedly made by drawing balls from the subject-designed urn and having robot players decide on whether to follow those recommendations in playing a game. Robots are known to follow recommendations only if there are no profitable deviations given knowledge of the recommendation device. Subjects are incentivised to design the devices whose recommendations robots always follow, that is the devices consistent with the correlated equilibria. We challenge subjects to find the correlated equilibria for such games as Prisoner’s Dilemma, Matching Pennies, Battle of the Sexes and two different versions of Chicken game. We compare subjects' performances under different games. We argue that our methodology of having subjects design recommendations devices provides a novel approach to evaluating the relevance of Nash versus correlated equilibrium solution concepts.
Для получения ссылки на Webinar-конференцию надо написать запрос на почту spb.economic.seminar@gmail.com с указанием имени и фамилии с рабочего адреса электронной почты.
Фото: Unsplash