This grant provides support for the expansion of a successful postdoctoral fellowship program run by the Council on Library and Information Resources (CLIR). The program aims to help academic libraries provide a new set of services that support data- and computation-intensive research through funding postdoctoral fellows devoted to data management and curation in the natural and social sciences. Grant funds will provide salary support to two cohorts of fellows (ten in 2013 and 12 in 2014) as well as various support and training activities such as professional training, travel, and networking with other data curation professionals.

Github is a new online service that helps programmers track and share their work. Based on "Git," a protocol for version tracking and the coordination of distributed contributions to software development, Github has become an extremely popular home for software projects both large and small, and has seen increasing use by scientists who develop software as part of their research. One notable feature of Github is its business model. There's no charge to set up an account and start posting, but there's a fee to keep your work private. This grant funds a research project by Jim Herbsleb of Carnegie Mellon's Institute for Software Research to evaluate how transparent "opt-out" development environments like Github affect the development of scientific software. Conducting case studies and analyzing archival data from Github, Herbsleb will investigate several key theses about the relationship between transparency and scientific software development, including how software developers use transparency to accomplish technical tasks, the role transparency plays in relationships between developers and the scientific community, and the difficulties transparent development environments pose for effective software development. Herbsleb's research has the potential to form the basis for policy recommendations on how transparency can be used most effectively to foster the development of scientific research.

Many researchers fail to appropriately capture, log, and version their work as it moves through the research process from data collection through multiple stages of cleaning and preparation to analysis. In part, this failure is due to the difficulty of logging changes in data as it moves from one software platform or set of scripts to another, each of which might be ideal for a particular part of the research process, but none of which are tied together by a common platform that can track the provenance of data as it moves from one system to the next. For decades, experimental scientists captured their research activities in lab notebooks. What is needed is a revitalization of that old idea: a lab notebook for the modern era of computational science. This two-year grant funds the development of just such an electronic lab notebook environment, called the IPython Notebook. Built on top of Python, R, and other widely used software languages in the data science community, the IPython Notebook is an early prototype computational platform that allows researchers to run a wide variety of high-powered data cleaning, modeling, and analysis algorithms inside a common computational environment. Grant funds will help IPython developers make the leap from early adoption to mainstream usage, focusing particularly on the development and scaling of features in three key areas: interactive exploration of data, collaborative authoring, and dissemination/sharing. Additional grant funds cover the salary of a full-time outreach coordinator to give presentations and tutorials at universities and professional society meetings, and funds to support the development of a set of live "notebooks" for use in introductory statistics classes, to better introduce students to the platform.

Shortly after the National Science Foundation began requiring all grant applications include a data management plan, a team based out of the California Digital Library developed and launched an online system to help researchers in the University of California system meet the new requirement. Named the "DMP Tool," the system contains information on funder data management requirements and on the data management resources available at participating universities, enabling researchers to quickly sketch a basic data management plan tailored to their particular proposal and institution. The system was a success. Within its first year, the DMP Tool was used to generate thousands of data management plans and has become an important resource for researchers. Because the DMP Tool was built under significant time constraints, however, the technical architecture that powers it was not developed with an eye towards expansion. As designed, the system is not prepared to accommodate the rapidly expanding number of funding agencies who have data management requirements, the increasing complexity of those requirements, or the quickly changing data management capabilities of member universities. The DMP Tool needs a core code re-write to build in the flexibility needed to meet rising demand. Funds from this grant provide support for a substantial rewriting of the DMP Tool software, with an eye toward flexibility and facilitating the effective use of the DMP Tool at a larger number of research institutions. The resulting website will better structure the metadata about research encoded in data management plans and offer broad analytics about research data management across funders, by capturing data management plans upstream of submission. Suitably rewritten, the DMP Tool has the opportunity to become the standard U.S. facilitator of data management plan creation.

This grant supports the continued development and expansion of "Software Carpentry", an initiative launched by the Mozilla Foundation to train scientists in best practices for how to use and develop software. Starting with immersive workshops, and following up with online resources and tutorials, the Software Carpentry project provides intensive, hands-on training that allows scientists to thrive in a research environment that is increasingly software-driven. Funds support Mozilla's efforts to expand the Software Carpentry program, conducting workshops and training scientists to design and lead workshops of their own. Additional funds will support the development and launch of a "Webmaking Science Lab", an online portal and set of complementary resources aimed at facilitating the open-source, collaborative, researcher-driven development of scientific software.