Research Data Horror Stories at 38C3
The Research Data Horror Stories workshop at the 38th Chaos Communication Congress (38C3) was an intense, enlightening, and sometimes chilling experience.
With a packed room of around 80 participants (despite only 60 chairs), the event highlichted the critical yet often underestimated importance of research data and software management (RDM).
The discussions revealed a plethora of horror stories from research data management and software development. Here's a recap of this unique session's key moments and learnings:
Horror Story Highlights:
Data Encryption Nightmare: A research lab experienced non-consensual electronic lab notebook data encryption. The recovery process spanned four months, leaving scientists with unlabeled data and irretrievable physical experiments.
Lesson Learned: Physical backupsand must be prioritized, and awareness of such risks should be integrated into training programs.Short-Term Projects and Long-Term Losses: The Wissenschaftszeitvertragsgesetz (German academic contract law) and short-term infrastructure projects led to significant criticism of NFDI and similar initiatives. This legal and systemic setup promotes a transient workforce, leading to lost knowledge and inaccessible data, exposing the unsustainability of current practices.
Lesson Learned: Sustainable funding and long-term perspectives are necessary for effective RDM and RSM.Faulty Software and Unchecked Code: A mathematician lost nine months of research due to flawed, inadequately reviewed software. Similarly, psychologists encountered issues caused by erroneous R scripts, underscoring the impact of insufficient coding skills.
Lesson Learned: Better testing and coding skills are needed to prevent false code and results.
Brainstorming Solutions - Highlights
The workshop was not just about lamenting challenges but also brainstorming solutions. Participants presented practical and actionable ideas to tackle systemic problems in research data management and software development. Here are some of the most notable suggestions:
Better Error Culture: Scientists should normalize admitting mistakes and actively work toward rectifying them. Journals and publishers must allow for minor corrections in papers and software, which will enhance credibility and transparency.
Improved Software Documentation: Documentation needs to be accessible, up-to-date, and comprehensive, especially in situations where urgent problem-solving is required.
Incentives for Code-Testing and Reproduction: Testing and reviewing code and data should become mandatory criteria for publication.
Public Funding, Public Code: Publicly funded research projects must mandate the use of open-source software, fostering collaboration and ensuring long-term sustainability.
Interdisciplinary Collaboration: Encouraging close collaboration between software engineers and researchers in other fields can bridge knowledge gaps and improve both coding practices and research outcomes.
Systemic Changes: Long-term contracts, better working conditions, and fewer but more impactful publications were proposed to strengthen knowledge transfer and promote a healthier academic environment.
The horror stories shared provided valuable real-world insights that can directly inform NFDIxCS personas and use cases. By understanding both the human and technical challenges faced by researchers, we can design tools and workflows that truly address their needs.
Thanks to the volunteers who helped facilitate the session and everyone who contributed their stories and solutions. While the workshop revealed numerous challenges, it also demonstrated the community's creativity, resilience, and commitment to improvement.
