Six things funders can do to maximise return on investment in software development

Posted by n.chuehong on 18 November 2013 - 4:59am

SalFalko_percent_cc-by-nc.jpgBy David De Roure, Carole Goble, Mark Parsons and Neil Chue Hong.

There is growing recognition that software underpins our research and infrastructure, and we are increasingly funding the use and development of software in research projects. Software, like data, outlives projects and hardware investments – it underpins the rigour of research and demands at least as much attention as data. It is important that investment in software is backed by simple changes in practice to enable maximum return.

This is even more evident when reading the UK House of Lords Science and Technology Committee report on Scientific Infrastructure which recommends "that the training and other costs, as well as the value of the skilled workforce needed to operate scientific infrastructure, are fully taken into account in developing the strategy and an underpinning investment plan [by the UK government]". A recent workshop on practices that lead to sustainable scientific and research software has confirmed our prior experiences in understanding best practice for software development, and the role of funders in this.

Here are six steps that we believe funders can take to ensure the quality and sustainability of this software:

  1. Invest in software training as part of grants (cf media training)
  2. Extend data management plans to include software (or introduce separate software management plan)
  3. Credit software reuse (monitored through the reporting of software production and reuse in research outcomes)
  4. Support a scheme for certification or kite-marking of research software (the Software Sustainability Institute can help with this)
  5. Insist that both code and data are published with papers (and encourage publishers to include this in peer review)
  6. Add software sustainability to the strategic plans of every research council, with an accompanying operational plan

Each of the first five points requires guidelines to be published and disseminated, but these are not novel or difficult concepts - there is existing best practice.

Funders have the ability to define policy in areas such as the inclusion of software in data management plans, but it is important that these are enforced. By ensuring that the six points listed above are supported, it incentivises researchers to take up best practice that will lead to an upskilled researcher base, and a greater return on investment in software.