Online seminar PIDs for Software

There was also the opportunity to exchange ideas in three groups during the event. In an interactive part, three questions were formulated for the participants to answer and discuss.
The following is a brief summary.

Research software faces significant hurdles in reproducibility, long-term usability, and recognition. Key challenges include:

• Maturity & Policy: Software is often developed quickly for specific projects with limited documentation, standards, and a lack of institutional incentives for proper citation or publication.
• Publication & Sharing: Inconsistent workflows, missing identifiers & metadata, and dependency issues complicate software publication and reuse.
• Resource Constraints: Limited time & resources hinder the implementation of good software development and publication practices.
• Technical Barriers: Lack of expertise, compatibility issues, and difficulties with hosting/sharing code (e.g., using platforms like GitHub) contribute to problems.
• Data & Documentation: Decisions about integrating datasets and a consistent lack of time for thorough documentation further complicate usability and reproducibility.

Workshop participants identified several challenges to adopting PIDs for research software. Key concerns center around workflow integration, metadata management, and long-term accessibility.

• Workflow & Tooling: Integrating PIDs into existing workflows is complex, particularly with manual publishing processes. Participants highlighted the need for convenient tools that fit specific needs.
• Metadata Synchronization: A “chicken-egg problem” exists regarding syncing PIDs with comprehensive metadata, requiring careful planning of who creates and manages it. The choice of metadata format also impacts usability.
• Version Control & DOIs: Existing version control systems (like GitLab) offer stable URLs, lessening the perceived benefit of additional PID mechanisms. Decisions around version-specific vs. latest-version DOIs require consideration.
• Accessibility & Ownership: Ensuring long-term software accessibility is a concern, especially when developers and publishers differ, or software is embedded within larger publications.
• Emerging Issues: Participants raised concerns about PID referencing for AI-generated software due to copyright and valuation questions.

Participants identified a comprehensive set of documentation elements crucial for ensuring software reusability. Key areas include:

• Licensing & Restrictions: Clear licensing information (adhering to software.reuse standards) and a defined scope outlining what the software can and cannot do.
• Technical Requirements: Detailed specifications for the software stack, operating systems, dependencies (with version numbers), and necessary resources.
• Usage & Execution: Comprehensive guides for installation, execution (including build/compile instructions), and minimal usage examples with test data.
• Context & Purpose: A clear description of the software’s scope, purpose, the problems it solves, and its FAIR indicators.
• Contribution & Support: Information on how to contribute, mechanisms for feedback, and contact information for authors/contributors.
• Dependencies & Citation: Clear citation information and details on all software dependencies.

Considering the identified challenges in reproducibility, usability, and long-term viability, participants advocate for more user-focused solutions and enhancements to clearly demonstrate the value of PID adoption within existing research software workflows. They emphasized that thorough documentation – essential for users, funders, and contributors – is a cornerstone of this effort, ensuring software remains reusable and impactful over time.

At the end of the event, additional questions were formulated in a Zoom survey that should help the PID Network Germany project to once again record the requirements and conditions for the use of PIDs for software in a structured manner. These results also underline the findings from the group work. More than 70% of participants stated that they find guidelines or policies for the use of research software very helpful. Furthermore, these respondents agreed that binding requirements (e.g. from third-party funders) could support PID use and implementation. In line with previous discussions, most participants cited a lack of awareness or training and the challenge of integrating PIDs into existing workflows as the biggest hurdles. At the same time, it is pointed out that sustainable funding for the implementation and maintenance of PID must be ensured.

We would like to thank all participants and speakers for the exchange.