Designing Knowledge-Driven Innovation Contests

Innovation contests have been studied primarily as mechanisms to obtain extreme valued solutions. In the current work, we propose that this conceptualization of innovation contests needs to be expanded to also consider the long-term benefits from the knowledge/capabilities generated by the contestants' efforts. We develop a novel framework that explicitly includes the generation and utilization of knowledge by participants in an innovation contest, and we analyze the implications of this framework for the design of serial contests. The model conceptualizes knowledge as arising from one of two modes of learning - Learning Before Doing and Learning While Doing. Using a model where contestants optimally allocate effort between execution and explicit knowledge generation, we show that the efforts expended by the contestant in a contest depends to a large degree on future rewards, especially so when learning emerges as a ``side effect'' of execution (Learning While Doing). Thus, contestants will exert effort in the earlier contests even when the associated reward is zero. We characterize optimal serial contest incentives and demonstrate that whether one should assign higher reward to the earlier or later contest depends on the mode and ease of learning, the participant pool's ex-ante knowledge, and the transferability of learning from one contest to the next