BCS calls for computer coding in scientific research to be more professional

The BCS has deplored a lack of professionalism in software development in medical science. This, says the society in a new policy paper, has led to a politicisation of modelling in the current Covid-19 health crisis.

Bill Mitchell, director of policy at BCS, The Chartered Institute for IT, said in a statement that accompanied the paper, Professionalising software development in scientific research: “The politicisation of the role of computer coding in epidemiology has made it obvious that our understanding and use of science relies as much on the underlying code as on the underlying research.”

In a statement to Computer Weekly, Mitchell said that by “politicisation”, the BCS means “that computer code used for scientific research, particularly around Covid-19, has been subject to a range of high-profile attacks that are not based on a rigorous analysis by independent software development experts, but seem to be designed to support a political position that is trying to undermine the science”.

He added: “Having an agreed, independent, open standard for coding in epidemiology would surely help protect the reputation of work that has significant social implications.”

In the discussion document, the institute said: “We welcome the government’s commitment to following science in developing policy responses to the coronavirus pandemic. We support the use of computational modelling in exploring possible outcomes of policy decisions, such as investigating which lockdown measures are likely to have the greatest public health benefits.

“At the same time, we consider that, at present, the quality of the software implementations of scientific models appear to rely too much on the individual coding practices of the scientists (who are not computer scientists) who develop them, rather than professional software development practices.”

To back up the BCS’s position, Mitchell cited a recent article in Nature, Three pitfalls to avoid in machine learning, which reported that “many machine-learning papers fail to perform an adequate set of experiments”, which has led to poor-quality research being published.

In its discussion document, the institute argues that the lack of widely accepted software development standards in scientific research has allowed for the “politicised” undermining of confidence in computational modelling, including in high-profile models informing Covid-19 policy.

It also calls for professional software development standards to be adopted for research that has a critical impact on society, such as criminal justice and climate change, as well as health.

The underlying code should also be made open source, says the BCS.

It calls on “those scientists who are working so hard and doing vitally important work in the current circumstances to engage with the relevant computer science specialists to ensure that they receive all of the benefits that best-practice software development can provide to their work”.

The institute said it would initiate a discussion about how to professionalise software development in scientific research among other bodies, including the Centre for Data Ethics and Innovation, the Alan Turing Institute, the Safety Critical Systems Club, the British Insurance Association, the Royal Society, the Royal Academy of Engineering, the Cabinet Office, NSHX, UKRI, and Public Health England.