All scientific fields report their results in publicly published and permanently archived papers. In a few fields, such as genetic sequencing, the community expects researchers also to publish their data sets, in a standard format, in a permanent archive. This project explores the next step in scientific publishing: capturing and publishing entire experiments that are fully encapsulated, ready for immediate replay, and open to inspection. Experiments and results could be examined, repeated, extended, and reused-- by anyone.
This planning grant focuses on assessing the breadth and depth of various communities' interest in such "open community archives of replayable experiments," and their requirements, feasibility, and key design issues. Such "active libraries" would directly connect published research to its sources in ways that support new analyses and new executions of computer-based systems. These archives would directly benefit many research and education communities, and their construction would involve a wide array of systems research challenges. A long-term goal is to build a coalition of stakeholders that includes computer and computational scientists, educators, librarians, publishers, and professional societies.
The intellectual merit of this project includes assessing many challenging issues, including encapsulation, virtualization, large-scale data management, user interface, and scientific culture and trends. In broader impact, such archives have the potential to change the process of scientific publishing for much computation-based research.
