Crowdsourcing with all-pay auctions: A field experiment on Taskcn

We investigate the effects of various design features of all-pay auction crowdsourcing sites by conducting a fieldexperiment on Taskcn, one of the largest crowdsourcingsites in China where all-pay auction mechanisms are used. Specifically, we study the effects of price, reserve price inthe form of the early entry of high-quality answers (shillanswers), and reputation systems on answer quantity andquality by posting translation and programming tasks on Taskcn. We find significant price effects on both thenumber of submissions and answer quality, and that taskswith shill answers have pronounced lower answer quality, which are consistent with our theoretical predictions. Inaddition, monetary incentives and the existence of shillanswers have different effects on users with differingexperience and expertise levels.