Australian researchers produced a great deal of original software as part of Australian Research Council funded grants, but it is hard to find – let alone know to use it for other projects.
Eva Maxfield Brown and Nic Weber used machine learning tools to estimate that 47% of ARC funded projects 2010-2019 created software infrastructure, tools or “code of some kind.”
Creation is not solely in STEM – 36% of economics, 24% of humanities and 15% of education projects used original software. However the big users, grouped by ARC research evaluation, are in the hard sciences: “technology” 72%; maths 68% and physical, chemical and earth sciences, 58%.
So, what is to be done?
The authors propose:
- software production descriptions in grant applications to “help ensure that research software work is not invisible labour nor redundant with previous efforts;”
- software is reported as a grant outcome: allowing re-use of “potentially valuable” ARC-funded products;
- a national census, “linking more ARC-funded research outcomes together such as datasets and publications; and
- identification of workforce needs in research software engineering and support.
Regardless of what occurs to improve awareness, the authors recommend that access should be open to all. Maxfield Brown and Weber found development occurred across grant sizes and university groups.
