Problem Specification in Crowdsourcing Contests: A Natural Experiment

Problem specification is a key aspect in crowdsourcing contests through which seekers convey their requirement and taste for the desired submissions. Hence, it is important to understand how problem specification should be framed to achieve better crowdsourcing contest outcomes. In this empirical study, we investigate the effects of a relatively more structured problem specification on contest quantity, solver quality, and idea quality. We leverage on a natural experiment setup on a major crowdsourcing contest platform where the problem specification of logo design contests changed from open-ended to structured. Our results show that the specification change impacts both seekers and solvers. Specifically, the number of contests increases after the change but the efficacy of top solvers and idea quality in the respective contests tend to be lower. We discuss the theoretical and practical contributions of this research.