The first step towards reproducibility of experiments based on software is to be able to actually find the software that was used in the experiment. While this seems to be self evident, in practice it turns out to be not easy at all, and the tools which are commonly used today have severe limitations.

In this talk, Roberto Di Cosmo will provide an overview of the issues we face when trying to ensure that digital content in general, and software in particular, is properly available and traceable. In order to address these issues, the Software Heritage initiative was started at Inria, and we will report here on its progress.