When the exciting evidence for gravitational waves was announced earlier this month, the accompanying paper from the Laser Interferometer Gravitational-Wave Observatory (LIGO) listed more than 1,000 authors from around the world. But like many important scientific projects, there was an unsung hero behind the decades-long research: software. While the LIGO detector facilities in Louisiana and Washington actually heard the “chirp” from two colliding black holes, the code written by scientists to simulate, collect, and analyze the signal were just as critical to the breakthrough.
Nonetheless, the traditional practices of scientific publication have made it difficult to properly credit software and the developers that built it. Journal articles are generally written to present scientific findings and share results, not code, a credit gap that has pushed many promising scientific programmers onto other career paths. But last year, a group of scientists including CI Senior Fellow and Argonne computer scientist Kate Keahey launched a new kind of journal, SoftwareX, to treat these contributions as valuable scientific instruments and give them the credit they deserve.
“When you create a scientific instrument, the science it enables very often exceeds a single scientific discovery by itself,” Keahey said. “It is therefore a paradox that software is not as well recognized and not as well appreciated as the scientific discoveries themselves.”
After less than a year of existence, SoftwareX recently received a rare honor: the Award for Innovation in Journal Publishing from the Professional & Scholarly Publishing Association of American Publishers. The award is only given out in years when the judges deem a journal sufficiently cutting-edge to earn the distinction.
Keahey says she’s most excited about the volume of submissions the young publication has already received since his first issue was published last summer. The enthusiasm validates the creator’s drive to highlight an underappreciated side of science, she said.
“We’re actually using fairly traditional publishing methods to do something new,” Keahey said. “Our main concern is to give credit to creators of scientific software and help articulate career paths for scientific developers. We have a lot of very talented people excited about working in science who don’t get the appreciation they deserve and don’t have well defined career milestones”
Despite its place in the well-established Elsevier science catalog -- which helps make its contents easily discoverable through traditional science search engines -- editors-in-chief Keahey, Frank Seinstra, and David Wallom made several innovative decisions while planning the journal. Instead of focusing on just one discipline, they chose to accept submissions from across fields, in order to promote sharing of software and avoid redundant development. For instance, an image analysis algorithm used to detect distant galaxies might be adaptable for finding tumors in medical images.
Unlike previous journals that published scientific software, they also chose to review submissions on the basis of scientific impact, instead of technical quality.
“If a piece of software has contributed to significant scientific discoveries or is used by thousands of users you can safely assume that it will compile and run," Keahey said.
They also developed a new publication format, called the Original Software Publication, or OSP, to preserve software at different stages of development (as explained in the video below). The format carries over some of the version control and collaborative potential of Github, but with the additional trustworthiness afforded by peer review (the journal also maintains its own Github repo). Instead of publishing just code and documentation, the editors also ask for a brief, descriptive paper that focuses on the impact of the software: how it helped advance science, and why researchers should use it.
Thus far, early acceptances have included anyFish 2.0, a platform to create models of fish behaviors, and Skinware 2.0, “a real-time middleware for robot skin.” In less than a year, the journal has received 120 submissions, and seen articles garner over 5000 downloads. For more on the journal, read an article at the Science 2.0 blog.