Solver capacity utilization and allocation on crowdsourcing platforms: an experimental study

We study innovation contests on crowdsourcing platforms that seek solutions to a set of problems from solvers who face capacity constraints in their solution-development efforts due to limited (financial, time, cognitive) resources. We analyze how solvers utilize their limited capacity and allocate it when competing in multiple contests by considering the moderating effects of solver uncertainty and platform growth. We build contest theory based on a game-theoretic model where a solver’s likelihood of winning a contest is determined by the quality of her solution, which improves with her effort and is also influenced by some output uncertainty. We show that solvers increase their capacity utilization when they face lower uncertainty or compete in a larger number of contests. Furthermore, when competing in multiple contests, solvers allocate their capacity evenly across all contests. More importantly, platform growth can improve the per-contest outcome if and only if the solver uncertainty is above a certain threshold. We test these theoretical predictions with controlled laboratory experiments by varying uncertainty levels and the number of contests. Our experimental findings show that solvers utilize less capacity than predicted in all treatments, but they utilize capacity better and allocate it unevenly in a multi-contest setting. Because of these effects, the per-contest outcome is better in a multi-contest setting than in a single-contest setting, even when theory predicts equal outcomes.