Události

Út 14.06.2016

Prof. Dražen Prelec (MIT Sloan) “Eliciting and Aggregating Information when Ground Truth is Unverifiable: The Meta-Prediction Approach”

Prof. Dražen Prelec

Sloan School of Management, MIT, Cambridge, Massachusetts, USA

Author: Dražen Prelec (Principal collaborators: H. Sebastian Seung, John McCoy)

Abstract: The twin problems of eliciting and aggregating information arise at many levels, including the social (wisdom-of-the-crowd) and the neural (ensemble voting). The first, elicitation problem involves crafting incentives that ensure that incoming signals are honestly reported; the second, aggregation problem involves selecting the best value from a distribution of reported values (Galton famously proposed the ‘democratic’ median, back in 1907). I will present recent work on elicitation (Bayesian truth serum) and aggregation of individual judgments, in a setting where ground truth is non-verifiable and prior information about individual competence is lacking. Non-verifiable domains encompass legal and artistic judgments, futuristic forecasts, as well as phenomenological “first person” descriptions of mental activity. The formal results use Bayesian game theory, and exploit information contained in predictions by individuals about how others will respond. Experimental results illustrate the approach.