Research software as an enabler and research output

This presentation was prepared and delivered at the virtual National Science and Technology Forum's Discussion Forum on Research Software on 7 October 2026. Abstract Research increasingly depends on software, yet much of that software is developed by researchers whose primary training is not in software engineering. This keynote explores what counts as research software, the many roles it plays across the research lifecycle, and how it can be both a research output and an enabler of further research and teaching. Drawing on examples from South Africa and the growing international Research Software Engineering community, we consider what it takes to move from code that works for an immediate research question to software that remains understandable, maintainable, and useful beyond its original context. As AI makes it easier to produce code, the need for sound engineering judgement becomes more, not less, important. Ultimately, good research software depends on bringing research expertise and software engineering expertise together. This upload contains the following files: Slides: NSTF-keynotes-slides.pdf The original, image-dense presentation Narrative / approximate transcript: NSTF-keynote-narrative.pdf Provides the spoken narrative and context that is not evident from the slides alone Level 3 accessibility companion: NSTF-keynote-accessibility-document-for-slides.html Created following the Equersa accessible-presentation guidelines to make the visual content accessible to visually impaired readers.