Competition for priority harms the reliability of science, but reforms can help

Incentives for priority of discovery are hypothesized to harm scientific reliability. Here, we evaluate this hypothesis by developing an evolutionary agent-based model of a competitive scientific process. We find that rewarding priority of discovery causes populations to culturally evolve towards conducting research with smaller samples. This reduces research reliability and the information-value of the average study. Increased startup costs for setting up single studies and increased payoffs for secondary results (a.k.a. “scoop protection”) attenuate the negative effects of competition. Further, large rewards for negative results promote the evolution of smaller sample sizes. Our results confirm the logical coherence of “scoop protection” reforms at several journals. Our results also imply that reforms to increase scientific efficiency, such as rapid journal turnaround times, may produce collateral damage by incentivizing lower-quality research; in contrast, reforms that increase startup costs, such as preregistration and registered reports, generate incentives for higher-quality research.