The scientific enterprise has long been based on the presumption of replication, although scientists have recently become aware of various corruptions of the enterprise that have hurt replicability. In this article, we begin by considering three illustrations of research paradigms that have all been subject to intense scrutiny through replications and theoretical concerns. The three paradigms are one for which the corpus of research points to a real finding, one for which the corpus of research points to a significantly attenuated effect, and one for which the debate is ongoing. We then discuss what scientists can learn—and how science can be improved—through replications more generally. From there, we discuss what we believe needs to be done to improve scientific inquiry with regard to replication moving forward. Finally, we conclude by providing readers with several different approaches to replication and how these approaches progress science. The approaches discussed include multilab replications of many effects, multilab replications of specific effects, adversarial collaborations, and stand-alone applications.

长期以来，科学企业一直基于可复制性的假设，尽管科学家们最近意识到企业的各种损害可复制性的腐败。在本文中，我们首先考虑三个研究范式的例子，这些范式都通过复制和理论问题受到严格审查。这三种范式是研究语料库指向真实发现的一种范式，研究语料库指向显着减弱效果的一种范式，以及正在进行辩论的一种范式。然后，我们讨论科学家可以通过更普遍的复制来学习什么——以及如何改进科学。从那里，我们讨论了我们认为需要做些什么来改进关于复制的科学探究。最后，最后，我们为读者提供了几种不同的复制方法以及这些方法如何推动科学发展。讨论的方法包括许多效果的多实验室复制、特定效果的多实验室复制、对抗性协作和独立应用程序。