Increasing the effectiveness and efficiency of research software development is both necessary and possible

Welcome to the third post in our six-part series exploring the future of research software in Germany. In each installment, we present opinion statements on why a national research software institution could and should be created to strengthen Germany’s research ecosystem.

The current heterogeneous and fragmented state of the German research software ecosystem results in considerable hidden costs and wasted resources. Numerous parallel structures exist, sometimes even within the same institution, which lead to redundant developments and inefficient, non-standardised custom-built solutions.

Additionally, keeping up to date with diverse, parallel projects and initiatives, let alone understanding them, requires time and resources. The present situation has now become so complex and diverse that the strict competitive logic of the German science system can no longer be the only driver and needs to be complemented by a logic of impact.

There is a more effective and efficient way forward: Local units or departments could benefit from operating under a larger umbrella without sacrificing their autonomy. Although community-driven, decentralised approaches to research software engineering are important, establishing individual groups within each faculty or institute would be costly and disadvantage smaller, less well-resourced institutions. Also, centralising certain functions, such as higher degrees of networking and coordination, could help create more cost-efficient and effective research software development. Even though it does not always pay off in the short term, the development of joint research software could foster stronger collaboration by establishing interdisciplinary structures. Thus, improving intra- and inter-organisational collaboration is particularly promising.

Local transfer approaches often result in limited user groups. A central unit or department could facilitate collaboration and coordination between different domain research groups, as well as between these groups and technical research or support staff. More central consolidation could ensure everyone has access to all available knowledge and support the development and establishment of shared quality standards. Clear and transparent project governance, even if minimal, can help build trust in the project and its ability to evolve over time. Although this restructuring poses challenges, particularly with regard to the redistribution of resources, greater structure and governance could help achieve the necessary effectiveness and efficiency improvements. However, the tension between these gains and institutional independence must be considered.

Another way to increase efficiency would be to better leverage the use of existing high-quality software. However, funders must recognise the value of supporting existing software through maintenance and use, and acknowledge the benefits of making strategic design decisions that require a higher initial investment but offer highly advantageous long-term effects on a software project’s evolution. This would enable research funds to be allocated more strategically to the continued development of existing relevant software, as well as new software with a good chance of long-term success. In this regard, encouraging the publication of research software with an open source software licence can make a software product less likely to become abandoned. Currently, an open sharing practice is by no means the standard case.

A national research software institution can serve as a consolidation hub, making knowledge available centrally, and helping to avoid duplication of efforts. By providing coordinated structures and governance, it can establish best practices and contribute to significant efficiency gains throughout the scientific community while preserving local autonomy. To funding agencies and policymakers, it can serve as the unified voice of the community advocating for necessary change.