Speaker : Dr. Saptarshi Ghosh

Abstract :

Manipulating Private Information to Evaluate Expert Talent Nejat Anbarci Saptarshi P. Ghosh Jaideep Roy Abstract An evaluator wishes to accurately evaluate an expert’s talent for obtaining and analysing information in prediction tasks. As part of an active talent-spotting strategy, he may interfere with the expert’s private signal by jamming or boosting its precision. We show that when the priors are significantly uninformative so that the prediction task is a priori difficult, the evaluator finds it optimal to jam the precision of the expert’s signal, while when the priors are significantly informative, making the prediction a priori easy, he boosts it. These results hold irrespective of whether the expert is a careerist or purely outcome driven.