Power of Bonus in Pricing for Crowdsourcing

We consider a simple form of pricing for crowdsourcing system, where policy is published priori, and workers then decide their task acceptance. Such widely adopted in practice its simplicity, e.g., Amazon Mechanical Turk, although additional sophistication to rule can enhance budget efficiency. With the goal designing efficient rules, we study impact following two design features policies: (i) personalization tailoring worker-by-worker (ii) bonus payment qualified completion. In Bayesian setting, only prior distribution workers' profiles available, first Price Agnosticism (PoA) that quantifies utility gap between personalized common policies. show PoA bounded within constant factor under some mild conditions, essential pricing. These analytic results imply complex be replaced by once it equipped with proper payment. To provide insights on pricing, mechanisms several profile regimes which may exist practice. primitive experiments support our analytical findings.