Synthesizing Winning Strategies: What Differentiates Experienced Solvers in Crowdsourcing Markets?

Mikhail Lysyakov, Siva Viswanathan

Abstract: Description Prior studies of online crowdsourcing platforms have examined participants’ behaviors and found that experienced solvers strategically time their submissions and are more likely to win a contest. But what gives experienced solvers an edge in these contests is not well understood. Our study seeks to understand what differentiates experienced solvers, with a particular focus on how they leverage information in open design contests. We use large-scale empirical analysis employing deep-learning algorithms and find that, while experienced solvers are similar to less-experienced solvers in a number of ways, experienced solvers are more adept at integrating information from prior highly-rated submissions from other solvers in a contest. We find that experienced solvers whose submissions are closer in similarity to a synthesized image of highly-rated prior submissions, are more likely to win. Our findings provide new insights into the winning strategies of experienced solvers and have implications for the design of such markets.