The paper implies that less reproducible papers have a greater influence on science because they are more highly cited. But an alternate explanation suggests t

The paper implies that less reproducible papers have a greater influence on science because they are more highly cited. But an alternate explanation suggests the opposite -- less reproducible papers are more highly cited because people are publishing papers pointing out the results are false. It is also quite telling that the biggest differences in citation counts are for papers published in Nature and Science. But in discipline specific journals (Figs. 1B,C), the effect is very modest. Practicing scientists know that Science and Nature are publish the least reproducible results, in part because they like "sexy" (surprising, less likely to be correct) science, and in part because they provide almost no detail on how experiments were performed (no Materials and Methods). The implication of the paper is that less reproducible science has more impact than reproducible science. But we know this is wrong -- reproducible results persist, while incorrect results do not (we haven't heard much more about the organisms that use arsenic rather than phosphorus in their DNA -- https://science.sciencemag.org/content/332/6034/1163 https://science.sciencemag.org/content/332/6034/1163 )