Published: 5th August 2020
Editors: Jeremy Cohen (Imperial College London), Ilektra Christidi (University College London), Neil Chue Hong (Software Sustainability Institute, University of Edinburgh), Alejandra Gonzalez-Beltran (Science & Technology Facilities Council, UK Research & Innovation)
* The editors wish to acknowledge all participants in the RSLondonSouthEast 2020 discussion session, the outputs of which are highlighted in this blog post and the associated report. See the conclusions section and the full session write-up report for further details.
RSLondonSouthEast 2020 was the 2nd annual workshop of the Research Software London regional RSE community for London and South East of England. The one-day workshop, which took place in February 2020, was again held at The Royal Society in Central London, supported by UKRI-EPSRC and the Society of Research Software Engineering, and brought together nearly 100 participants. A key element of the day’s schedule was the discussion session which provided the attendees with an opportunity to discuss their thoughts and ideas around some specific aspects of RSE.
Photo by Kelly Sikkema on Unsplash
Wishlists for the future of RSE
The brief for the discussion session invited participants to form into small groups around 4 general areas of RSE and to build “wishlists” for the top 5 things they felt were most important or that they’d most like to see happening in these areas. The areas considered were motivated by the “Four Pillars of Research Software Engineering”, and were:
- Policy
- Training
- Community
- Software Development
In this post, we provide an overview of the outcomes of the session and highlights of the discussion and top “wishlist” items for each area. For further information, you can read the detailed write-up of the session which is now available on Zenodo.
Policy
This area has, ultimately, been behind the many developments in research software over recent years, from the coining of the name “Research Software Engineer” to support for the RSE Network, Fellowships and the growth in the number of groups at institutions within the UK and several other countries.
The topics discussed included the ability to request funds for longer-term software maintenance and sustainability aspects in grant applications, particularly where the software underpins wider research communities. Job stability was also highlighted with some participants keen to see a need for “directly allocated” RSE posts recognised. The recognition of RSE work also featured here, with a desire to see better recognition of RSE outputs in research assessment activities. Finally, group members were keen to see stronger guidance on the use of funding to support open source development work.
Training
Training represents a key part of work for many RSEs, whether that’s developing and delivering training material or attending courses to improve their technical skills and knowledge on software or the research domain(s) in which they work. Areas highlighted by the groups discussing training included providing opportunities for RSEs to grow their knowledge of areas beyond software development, for example, team and project management skills. A need to understand pedagogy and improve approaches to delivering training material was also considered important. Outreach was also highlighted in terms of developing better processes for knowledge exchange and building training links with non-academic institutions. In terms of technical topics, one area that was raised was developing software at scale and providing opportunities to train RSEs in areas such as GPU programming, machine learning algorithms and other aspects of development that are becoming much more common as data volumes and computational requirements continue to increase.
Community
RSE teams are now increasingly common at academic institutions but beyond these teams, RSEs tend to work alone or in very small teams embedded within research groups. Communities are important in ensuring that RSEs have opportunities to network, share technical knowledge and skills, and feel part of a group. The discussion groups covering this topic felt that they would like to see better convergence between RSE communities and more general academic and industry communities. Alongside this, a desire to strengthen outreach to the wider research community through mediums such as blogs, webinars, etc was also raised. They highlighted a need for communities with different focuses, perhaps through the setting up of sub-communities – the area of commercialisation for research software was specifically mentioned. Finally, there was interest in the idea of RSE community ambassadors who would help to support, promote and grow RSE communities, something that takes extensive effort.
Software Development
The final area considered – software development – is perhaps the most substantial aspect of research software engineering. Rather than focusing on specific software tools, languages and techniques, the discussion focused more on the principles of software development. This included suggestions that there need to be ways to make it easier to understand what software is required for a given task; supporting interoperability; and software design that simplifies research workflows/pipelines. The importance of open source software throughout software stacks was something that discussion participants were keen to highlight as well as the need to extend Integrated Development Environments to support research-related software development tasks. Finally, the need to support data checking and governance through specialist software tools was also raised.
Conclusions
It is clear from the outputs of the RSLondonSouthEast2020 discussion session that there are lots of opportunities in the research software space to address challenges and develop new capabilities that can help RSEs to achieve more with fewer resources. As a relatively new and rapidly developing area, Research Software Engineering continues to provide extensive scope for novel and interesting approaches that can help to improve the quality of research software and ultimately enhance the quality of the research that the software supports. The editors would like to credit all the participants in the discussion session with the outputs highlighted in this blog post, and a list of contributors can be found at the end of the full discussion session report.
Acknowledgements
The editors would like to thank UKRI-EPSRC who supported the RSLondonSouthEast 2020 workshop, which made it possible to hold the discussion session, through Jeremy Cohen’s RSE Fellowship (EP/R025460/1) and the Society of Research Software Engineering who sponsored the workshop’s poster session.