"Where do the people come from?"—A mixed methods investigation of Research Software Engineering

Lee, Graham

doi:10.5281/zenodo.17967403

"Where do the people come from?"—A mixed methods investigation of Research Software Engineering

Lee, Graham

Publisher: Zenodo
DOI: 10.5281/zenodo.17967403
Repository: zenodo.org
Language: en

Abstract

Over the last decade, the idea of a role of Research Software Engineer (rse) has gained some adoption. The RSE is a skilled software engineer who focuses on the construction of software for use in research. In a brief time, the idea has gone from a defining term created at a workshop on software sustainability to a job title and group name used at numerous research institutions worldwide, with member societies sharing experience and advocating for career recognition. In short, we are witnessing the creation of a new profession of Research Software Engineering.In April 2025, the presenter defended their doctoral thesis, in which we described their design and implementation of a mixed-methods investigation (containing both quantitative and qualitative components) into the development and propagation of RSE, drawing on relevant theories from the sociology of professions. We believe this to be the first such systematic, scholarly investigation into RSE.We implemented one quantitative and three qualitative studies of RSE, addressing questions on: the state of software engineering knowledge in research; the state of RSE as a profession; the independence of RSE from existing research practice; and the connection between the relationship of researchers and RSEs, and research software practice. Combining these studies into a single mixed-methods programme, we create an overall picture of the state of RSE one decade into its inception, and an evidence-based “manifesto” for future change in the field. We present our research into RSE, our findings, and our recommendations from the manifesto.A recording of this session is available on YouTube: https://youtu.be/81D5KFa-Oag

Full text

Graham Lee, September 2025 “Where do the people come from?” A mixed methods investigation of Research Software Engineering Graham Lee, September 2025 “I really don’t like Gaussian processes. I wanna write Python libraries.” A mixed methods investigation of Research Software Engineering in the U.K. Motivation To discover how Research Software Engineering can enable better research by aligning its practices with the things researchers value. Research Questions RQ1 Do UK-based researchers have the skills and training they need to fulfil the software-related requirements of their work?! RQ2 Does RSE exhibit the four properties of a profession in the functionalist theory of professions: a body of expert knowledge; technical autonomy; valuing service to others; and privileged status? ! RQ3 How do the institutions and logics of research influence the way an RSE performs their work, and what competing institutions vie for legitimacy? ! RQ4 Does the structure of the relationship between researchers and RSEs in a research group aﬀect the research software practice? Methodology A constructivist, mixed-methods approach 1. Survey of Oxford University researchers to quantify the software skills gap.# Helps identify the prevalence of informal learning and unmet training needs within the institution.! 2. Open-ended interviews with RSEs and other stakeholders.# Provides deep understanding of RSEs' self-perception, challenges, and aspirations regarding professional identity.! 3. Search and analysis of public blog posts about RSE.# Reveals how the RSE community articulates its value, practices, and navigates academic expectations in public discourse.! 4. Focus group discussions with researchers and RSEs working together.# Directly examines how collaboration models (e.g., embedded vs. pooled RSEs) influence software quality and community building. RQ1: The skills gap persists Finding: Most researchers need software skills and learn through self-teaching or peer mentoring, with formal training close behind but little RSE-led mentorship.# # Practical Implication: RSE has yet to eﬀectively close the skills gap for the broader research community. RQ2: A liminal, unrecognised identity Finding: RSE lacks a consistently defined body of knowledge, and its autonomy, service, and status are precarious, often perceived as “granted” by Principal Investigators. RSE is often perceived as a struggle for recognition against “traditional, snobbish academic cultures”.! Practical Implication: RSEs need to proactively define and articulate their unique value proposition and advocate for clearer career pathways and recognition, distinguishing themselves from technical roles that make less-direct research contributions. Picture credit: Paramount RQ3: RSE is isomorphic with academia Finding: RSE is highly dependent on academic structures (funding, career paths) and adopts academic logics to maintain legitimacy. Software engineering practices are often adopted incompletely and mimetically (copying visible practices) rather than systematically. Key SE areas like requirements management and software maintenance are consistently problematic or ignored.! Practical Implication: RSEs must strategically navigate academic politics, emphasising the tangible benefits of good software practice (reproducibility, impact) in terms that resonate with traditional academic incentives (publications, grants). This includes championing neglected areas like robust requirements and maintenance. Photo by Volodymyr Dobrovolskyy on Unsplash Recommendations A Manifesto for RSE change (1/2) •Funders:! •Systematically identify software that needs to be sustainable using a clear checklist.! •Ensure dedicated funding for ongoing maintenance for sustainable software.! •Research-producing organisations:! •Establish parallel career pathways and job descriptions for software roles.! •Incorporate software responsibility into codes of ethical practice.