Reproducibility as a service

Recent studies demonstrated that the reproducibility of previously published computational experiments is inadequate. Many these never recorded or preserved their environment, including packages installed in language, libraries on host system, and file locations. Researchers have created tools to help mitigate this problem, but assume experiment currently executes. Thus, do not facilitate large number experiments. This situation improving; researchers continue publish without using tools. We define a framework distinguish between actions taken by researcher presence environment enable reproduction an when has been lost clarify gap what existing are capable required reproduce The difference approaches lies availability environment. provide access original perform proactive reproducibility, while those who only retroactive reproducibility. present Reproducibility as Service (RaaS), which is, best our knowledge, first tool explicitly designed demonstrate how RaaS fixes many common errors found R scripts Harvard's Dataverse preserves recreated Finally, we discuss service such also helpful ‘artifact evaluation assistant’ journal's publication pipeline.