A new challenge for computational reproducibility emerges: code often becomes outdated quickly. By examining 20 studies from the ISPS Data Archive, we discovered nontrivial errors in seven of them.
Addressing the Problem:
- Researchers face difficulties maintaining old code due to rapid changes in programming languages and libraries.
- Current practices rely heavily on static code/data sharing without considering long-term viability.
Proposed Solution: Active Maintenance Plans:
We advocate for researchers committed to reproducibility to develop active maintenance plans. This involves anticipating future issues with software dependencies rather than just making the initial resources available.
Recommendations:
- Data archives should encourage and reward code maintenance efforts.
- Journals can incorporate requirements for sustainability into their open science policies.
This suggests that ensuring computational reproducibility requires more than passive sharing; it demands foresight and ongoing commitment from researchers.
