Diversity in Reproducibility

International audienceIn our previous contribution, we proposed computational modelling-related definitions for replicable, i.e., experiments within a model can be recreated using its original codebase, and reproducible, i.e., a model can be recreated based on its specification. We stressed the importance of specifications and of access to codebases. Furthermore, we highlighted an issue in scholarly communication — many journals do not require nor facilitate the sharing of code. In contrast, many third-party services have filled the gaps left by traditional publishers (e.g., Binder, 2016; GitHub, 2007; Open Science Framework, 2011; ReScience, 2015). Notwithstanding, journals and peers rarely request or expect use of such services. We ended by asking: are we ready to associate codebases with articles and are we prepared to ensure computational theories are well-specified and coherently implemented