Benign and malignant scientific irreproducibility

Irreproducibility is a widely recognized problem harming scientific progress. Here, we classify types of scientific irreproducibility based on deleteriousness using a cancer analogy: benign, pre-malignant and malignant. We discuss the difficulties of challenging potentially irreproducible papers, especially in high impact journals and/or by famous authors. We advocate for high impact journals publishing papers with large translation value, to implement mandatory author reflections on the reproducibility, utility and relevance of their work 5 to 10 years after publication. We suggest testing this idea with 100 papers published in elite journals between 2003 and 2008, to show the utility and need for these reflections. Knowing they will have to report on their results, will make researchers more careful and likely lead to a drop in scientific irreproducibility.