Building an Honest and Capable Crowd Workforce

One of the major challenges of crowdsourcing is to build a labor force that is not only capable but also honest. Prior research on crowdsourcing-mechanism design mostly focuses on motivating individual worker's effort, while neglecting the importance of building highly trustworthy crowd labor force who tends to honestly report the quality of the microwork. To fill this gap, we propose a novel mechanism in this paper. The proposed mechanism combines the principal-agent model and signaling game to enable the information exchange from crowd workers to a requester, and to provide dynamic financial incentives/punishments to honest/dishonest behaviors identified in random quality inspections. To evaluate our proposed mechanism, we perform extensive simulations. Our results suggest that the proposed mechanism is effective and efficient to motivate workers who are capable but less honest to change their behavior when sending signals to indicate their finished microwork, while the extra cost of implementing the proposed mechanism could be minimal. With the behavioral changes of the "liars", a highly capable and honest labor force of crowd workers may be developed. The model is also useful as a foundation for theoretical and empirical studies of information exchanges from worker to a requester in the crowdsourcing market.