New FAIR principles for research software are set to bolster research integrity and usher in welcome changes in scholarly practices, as they have for research data in recent years. The ARDC’s draft Research Software Agenda for Australia incorporates the principles with the aim to make research software a 1st-class research output.
Six years ago when the FAIR principles for data were being crafted, the authors fully intended that the principles should also be a good fit for research software.
This month, a set of draft principles for ensuring that research software is findable, accessible, interoperable and reusable (FAIR) has been reviewed by a working group comprising the Research Data Alliance (RDA), the Research Software Alliance (ReSA) and Force11 and the final principles are set to published imminently. ARDC team members participated in the working group to develop the draft FAIR principles for research software.
“FAIR is a cross cutting theme across all of ARDC’s programs,” says ARDC Software Program Manager Tom Honeyman.
“We’ve had great success with FAIR for data. Over the coming months, we’ll be looking at how we might adopt the FAIR principles for research software, and socialising them among the research community, alongside communicating approaches to software citation.”
A global team effort
Neil Chue Hong of the UK’s Software Sustainability Institute is leading the drafting of the principles. He credits the huge number of people who contributed voluntarily to the principles, under the guidance of an RDA working group.
“This work has really engaged the community worldwide,” says Neil. “We’ve had about 250 people giving use cases and opinions, and about 60 people have made hundreds of comments on drafts. I’m hopeful the [final] draft is pretty close to what we need.”
Australia has been a leading country in the effort, he adds, with the work group being led by Australian Michelle Barker from ReSA and a whole swag of Australian input, including from the ARDC.
“The level of engagement signals from the community a desire to tackle research software,” says Tom Honeyman. “So there’s interest. And we at the ARDC are right there, ready to take it forward, as part of a national research software agenda.”
Why does research software need to be FAIR?
Neil Chue Hong has been fundamental in the push for a set of FAIR principles for research software. Ten years ago, he co-founded SSI partly to address the finding that in the UK much of the software created through research funding was neither robust nor reusable.
“A lot of people go into research with a desire to improve society’s problems or they have an innate desire to focus deeply on a narrow topic,” says Neil.
“It’s been a long time since you could do that alone. You have to be able to build on the work of others and you can’t do that if things are not FAIR, doubly so for software. If I can build on something someone else has done, to get to the knowledge at the end of it, that gets me there faster.”
Tom Honeyman sees it as a way of boosting research integrity: “The principles in general are trying to usher in a change in practice in scholarly objects, getting people to make their data and associated objects available.”
The principles also come at a good time for Australia and other signatories to the OECD recommendation on access to research data created with public funding. Amended in 2021, the recommendation now calls for research software arising from public funding to be shared for greater public benefit.
Interoperability and reuse are different for software
Identifying where the principles for software need to differ from those for data has been a big chunk of the work to date. “It’s been a great opportunity to dig into what’s the same and what’s different and to think about the vision for software,” says Tom Honeyman.
The guiding principles for findability and accessibility are largely the same as those for data, except for versioning—software goes through more versions more quickly than data so versioning practices are quite mature, whereas it’s still early days for data versioning.
What it means to be interoperable and reusable has required some tweaking of the principles.
“For data to be reusable, it needs to be licensed and well described,” says Neil Chue Hong. “For software, there’s an additional element—can it be compiled and executed? Making the source code available is a great step but to make software reusable, we need to go beyond that and not just inspect it but actually run it.”
The key thing, Neil emphasises, is that we keep to the ethos of what the authors intended rather than the letter.
Adopting the principles
Defining the principles is only the first step and, as with all RDA recommendations, they will need to be adopted before they can be endorsed by RDA.
As well as the ARDC, in Europe the life sciences peak body Elixir, the Netherlands eScience Center, and the ZB MED Information Centre for Life Sciences in Germany are all looking to adopt the principles.
For the FAIR4RS working group, the next step is to provide the research sector with a roadmap for adopting the principles, then metrics, indicators and a way of certifying that a software object is FAIR.
“I think adoption is different for software,” says Neil. “For data, you work with a publishing venue and deposit your data. But software is often self-published, released through something like GitHub or a project website.”
Ultimately, he says, achieving adoption among researchers will come down to incentives and policies.
In a boost for the ARDC, Paula Andrea Martinez, a co-chair of the RDA FAIR4RS working group and ReSA Community Manager, has joined the software team at ARDC. Paula will now work with Tom Honeyman to take the initiative forward, and as part of her duties will continue to support ReSA as Community Manager one day a week.
Learn more about the Research Software Agenda for Australia.
Learn more about the FAIR4RS principles in this interview with Neil Chue Hong, in conversation with Mike Heroux (21’10).