Software development practices in academia: a case study comparison

TL;DR

This study examines the evolution of software development practices in four open-source scientific projects over nine years, revealing how team dynamics influence development strategies and scientific productivity.

Contribution

It provides a detailed case study analysis of academic software development, highlighting the impact of team transient nature on practices and productivity over time.

Findings

Team size influences development strategies

Time to scientific output is approximately three years

Team transience affects adoption of engineering practices

Abstract

Academic software development practices often differ from those of commercial development settings, yet only limited research has been conducted on assessing software development practises in academia. Here we present a case study of software development practices in four open-source scientific codes over a period of nine years, characterizing the evolution of their respective development teams, their scientific productivity, and the adoption (or discontinuation) of specific software engineering practises as the team size changes. We show that the transient nature of the development team results in the adoption of different development strategies. We relate measures of publication output to accumulated numbers of developers and find that for the projects considered the time-scale for returns on expended development effort is approximately three years. We discuss the implications of our findings for evaluating the performance of research software development, and in general any computationally oriented scientific project.