Tracking research software outputs in the UK

TL;DR

This paper investigates how UK research institutions store and share software outputs, revealing low sharing rates and issues with links, which threaten the long-term sustainability of research software in the scientific community.

Contribution

It provides an empirical analysis of UK research software registration and sharing practices using UKRI GtR metadata, highlighting gaps and challenges in research software dissemination.

Findings

Low proportion of software reported as research outcomes

High incidence of missing or erroneous URLs for software links

GitHub is the most common platform for publicly funded research software

Abstract

Research software is crucial in the research process and the growth of Open Science underscores the importance of accessing research artifacts, like data and code, raising traceability challenges among outputs. While it is a clear principle that research code, along with other essential outputs, should be recognised as artifacts of the research process, the how of this principle remains variable. This study examines where UK academic institutions store and register software as a unique research output, searching the UKRI's Gateway to Research (GtR) metadata for publicly funded research software in the UK. The quantity of software reported as research outcomes remains low in proportion to other categories. Artifact sharing appears low, with one-quarter of the reported software having no links and 45% having either a missing or erroneous URL. Of the valid URLs, we find the single largest category is Public Commercial Code Repository, with GitHub being the host of 18% of all publicly funded research software listed. These observations are contrasted with past findings from 2023 and finally, we discuss the lack of artifact sharing in UK research, with resulting implications for the maintenance and evolution of research software. Without dissemination, research software risks demotion to a transient artifact, useful only to meet short term research demands but ultimately lost to the broader enterprise of science.