Seven out of ten academics say they would not be able to do their work without research software, according to a survey published by the UK's Software Sustainability Institute (SSI) last month.
The survey, which included responses from over 400 researchers at 15 leading UK universities, found that 92% of academics use research software. A wide range of software is in use, with 566 different packages reported (the 'Wordle' (right) shows relative popularity of these).
In addition to software use, the survey also asked researchers about software development: 56% of respondents reported developing their own software, but over one fifth of these have no training in software development. "You wouldn't allow an untrained mechanic to work on your car, and I'd guess that you'd be pretty unlikely to visit an amateur surgeon," says Simon Hettrick, deputy director of the SSI. "Research software is often both incredibly complex and vital to the production of reliable results. Having a large number of researchers developing software with no training in how to perform that development is liable to lead to unreliable results."
While the results of the survey showed no difference between the percentage of men and the percentage of women who use scientific software, a large difference was reported in terms of development. 70% of male respondents reported developing their own research software, compared to only 30% for female researchers. This disparity is reflected in training: while 63% of men have received software development training, this is only the case for 39% of women.
Hettrick argues that there is a need for researchers to reassess their reliance on software, adding that it is often erroneously viewed as a disposable tool for generating results for publication. "This mindset leads researchers to spend less time on, and give less prominence to, their software than it needs," he says. Hettrick believes that the situation could be improved by giving researchers proper credit for the time they invest in producing good software.
Yet another problem highlighted by Hettrick is the lack of clear career paths for software engineers in academia. "It's difficult to hire them in academia, and there's no route for promotion so they're even more difficult to hold on to."
- Andrew Purcell