Reproducibility of Computational Results: Opening Code and Data
Scientific computation is emerging as absolutely central to the scientific method, but the prevalence of very relaxed practices is leading to a credibility crisis. Reproducible computational research, in which all details of computations — code and data — are made conveniently available to others, is a necessary response to this crisis. Questions emerge regarding scientists' incentives and motivations to share. This talk presents results from a survey of computational scientists to determine the factors that facilitate code and data sharing and those that create barriers. One major result finds that sharing is done for reasons other than direct personal gain, but when scientists choose not to reveal data or code this is due to perceived personal impact. A second major finding is the prominence of Intellectual Property concerns with regard to not sharing code and data. Solutions to the various barriers are discussed, including how the "Reproducible Research Standard" (Stodden 2008), which proposes a licensing structure consonant with scientific norms, can thus encourage open sharing in scientific research.