Software in science is ubiquitous yet overlooked

Citation

Hocquet, Alexandre and Wieber, Frédéric and Gramelsberger, Gabriele and Hinsen, Konrad and Diesmann, Markus and Pasquini Santos, Fernando and Landström, Catharina and Peters, Benjamin and Kasprowicz, Dawid and Borrelli, Arianna and Roth, Phillip and Lee, Clarissa Ai Ling and Olteanu, Alin and Böschen, Stefan (2024) Software in science is ubiquitous yet overlooked. Nature Computational Science. ISSN 2662-8457

Full text not available from this repository.

Abstract

Software is much more than just code. It is time to confront the complexity of licenses, uses, governance, infrastructure and other facets of software in science. Their influence is ubiquitous yet overlooked. In March 2020, Neil Ferguson, the scientist whose epidemiology model was used to justify COVID lockdown policies in the UK and around the world, was urged to make his model’s source code public. The model received some criticism on scientific grounds, but the most vocal objections targeted its software engineering aspects, calling it poorly designed, written and documented1. Such a culture clash is not surprising to some computational scientists, whose daily routine consists of designing, writing, maintaining, supporting, testing, debugging, adapting to new hardware, documenting, sharing, licensing and packaging a piece of software. Both computational researchers and software engineers are used to interacting with different temporalities, constraints, norms and work cultures. In June 2020, in the wake of Ferguson’s controversies, colleagues across the sciences and humanities published a timely and relevant manifesto in Nature that proposes “five ways to ensure that models serve society”2. Yet the manifesto does not mention the concept of software in their consideration of models. We believe this is lacking because models and software are entangled in science, and software does critical work that models cannot perform on their own. Software is indeed difficult to define, often being mistaken for code or algorithms. As historian of computing Thomas Haigh puts it: “Software always involves packaging disparate elements such as computer code, practices, algorithms, tacit knowledge, and intellectual property rights into an artifact suitable for dissemination”. Scientific software involves a diversity of practices regarding programming, governance, licensing, distribution, maintenance and support. It is developed and used across a myriad of scientific disciplines and programming traditions. It ranges in size from personal ‘scripts’ to huge projects involving entire communities and global infrastructure. It encompasses freely shared code as well as commercial packages. In this Comment, we emphasize the complexity of scientific software as a multifaceted socio-technical (and historically grown) system. We describe facets of software that we define as vantage points from which the different dimensions of software can be understood. The multifaceted nature of software implies that the work done by software has technical, legal, sociological and epistemic consequences. Models and software are entangled in computational science, and much remains to be done to comprehend these consequences. We also point out the diversity of situations involving software in computational science, which further complicates how to approach software facets. We highlight a few case studies, with the hope that this starting conversation about software will be enriched by further input.

Item Type: Article
Subjects: Q Science > QA Mathematics > QA150-272.5 Algebra
Divisions: Faculty of Creative Multimedia (FCM)
Depositing User: Ms Nurul Iqtiani Ahmad
Date Deposited: 01 Aug 2024 08:16
Last Modified: 01 Aug 2024 08:16
URII: http://shdl.mmu.edu.my/id/eprint/12733

Downloads

Downloads per month over past year

View ItemEdit (login required)