For too long software has been considered a third wheel, tagging along, taking up time and effort and resources from the “real” objective – scientific discovery. “It deserves better,” says Kate Keahey, senior computer scientist at Argonne National Laboratory.
The editorial traces the essential role that software has played – and continues to play – in what the authors call “the most significant scientific times: the detection of gravitational waves.”
Albert Einstein predicted gravitational waves in the early 1900s, but it was not until a century later that they were actually detected. The reason? Extremely sensitive instruments needed to be built, and “software was crucial from the very first design stage.”
But that was only the beginning. Software played a continuing role in gravitational wave detection – from instrument operation, to data collection, analysis, and signal modeling, and finally to release of the data to the broader scientific community.
Keahey and her coeditors use a historical account of the LIGO Scientific Collaboration to illustrate the challenges that such large and distributed software development efforts face. Should the data analysis software environment be centralized or decentralized? How can research software sustain its quality as a scientific collaboration grows? What about the need for software upgrades – does one rewrite or replace or simply “glue together” new formats and features? The individual articles of the special issues illustrate how those choices were made in practice in the context of specific discovery enabling tools.
At the end of the editorial, Keahey and her coeditors provide a brief assessment of why scientific software remains undervalued: “Software matured from an ingenious but entirely optional innovation to an indispensable tool of scientific discovery quickly and stealthily, and our perception and understanding of its now key role is slow to catch up.”
The editors hope this special issue will help change that perception and awaken the community to what needs to be done to recognize software not as a third wheel but as a critical component in scientific discovery.
The special issue, including the editorial by Keahey et al., is available online on the journal website.
For an interesting interview with Keahey, see the article in Elsevier Connect.