How the project list was defined and compiled
Definitions and distinctions
Tier 2 software (“Community or Shared Software”) includes software that is reused across multiple research initiatives. Tier 3 software (“Essential or Foundational Software”) is software that has become a shared resource relied upon by multiple institutions or disciplines. This software has a central role in the ecosystem, with broad adoption, essential in multiple research areas, and such with high social dependencies.
On the other hand, in our opinion, a research software infrastructure should be understood differently from tier 3 software. We define research software infrastructure as the shared technical and socio-organisational services, offerings and resources that enable the lifecycle management (e.g. the sustainable development, deployment, use, sharing, and maintenance) of research software (this is analogous to defining any other digital information infrastructure). These resources may include “infrastructure software”, such as code repositories, computing platforms, virtual research environments, continuous integration (CI) systems, as well as licensing frameworks, and governance models.
Due to their high social dependency, tier 3 software and infrastructure share the necessity for a distributed responsibility and shared stewardship. Often, multiple institutions co-fund development, there are multiple maintainers across organisations, and there is a collective steering of its roadmap. However, rather than being a single digital artefact or service, research software infrastructure denotes the underlying system of shared services, standards, and socio-organisational arrangements that enables the lifecycle management of such artefacts in the first place.
While the boundary between tier 3 and infrastructure seems clean in principle, in practice some tier 3 projects (e.g., a foundational library) may grow their own governance model, contribution rules, CI pipeline as part of the project. Thus, they start to produce infrastructure-like components internally which may lead to edge cases.
Methodology
The initial project list was generated with the assistance of ChatGPT in July 2026 and subsequently refined through an extended discussion and review process. ChatGPT evaluated the identified projects and proposed an initial classification as tier 1, tier 2, or tier 3 software.
This AI-generated classification was used only to identify promising tier 3 candidates and prioritize them for human review. It was not used to determine the final classification.
Two reviewers independently assessed the leading tier 3 candidates. A project was included in the filtered tier 3 list only when both reviewers agreed that it met the tier 3 classification criteria. Projects without agreement were not included in the filtered list.
The resulting list should be understood as a preliminary, prototypical selection rather than a complete or definitive classification. Further verification, correction, and community-informed curation are still required.
Publication goals
The publication aims to advance both conceptual and community-oriented work around tier 3 research software. A key goal is to refine the distinctions between tier 2 software, tier 3 software, and digital research infrastructure, thereby contributing to ongoing research in the RSE community and related fields.
At the same time, the publication seeks to increase the visibility of tier 3 software projects and the often insufficiently recognised research software development and maintenance work associated with them (“hiddenREF”). By making parts of the tier 3 software ecosystem more visible—including relevant stakeholders, Research Software Engineers (RSEs), Research Software Specialists, and maintainer groups—the publication is intended to contribute to a clearer understanding of the actors and structures involved.
Another goal is to actively involve the community in validating and further developing the preliminary project list. Structured feedback processes should enable corrections, additions, and other forms of community input, helping to improve both the classification approach and the representation of the wider tier 3 software ecosystem.
Finally, the publication is intended to create opportunities for deeper engagement with tier 3 project teams, for example through participatory workshops or in-depth interviews. Such engagement can help identify specific requirements for sustainable maintenance funding and gather practices and evidence that may inform future advocacy and support activities, including the showcasing, recognition, and long-term support of tier 3 research software.
Interactive comparison
Elements by broad domain
Loading chart data…
Filtered dataset
Tier 3 software list
Loading filtered list…