SC Colloquium: "Challenges in Enabling Reproducible and Reusable Computational Research"
Data Management Services
Johns Hopkins University
Being able to examine, reproduce, and extend the work of others are cornerstones of the scientific process. However, in many disciplines, opening up the data, executable software, and source code associated with research is not an accepted tradition. This means that an important portion of scholarly output is invisible. Even in computationally-oriented fields, where the ability to examine an implementation of an algorithm is highly important, code too often fails to work as advertised or is otherwise unavailable. This means that it becomes impractical, or in some cases, impossible to verify the implementation of an algorithm, or the result of an analysis. In addition, it results in inefficiencies in attempting to reuse research products because data may need to be re-collected and algorithms re-implemented. In this presentation, I will give an overview of some of the issues surrounding transparency, reproducibility, and reusability of computational research and what we can do to address them.