Identifying Socio-Technical Risks in Open-Source Software for Scholarly Communications: Tools, Metrics, and Opportunities for Libraries to Support Sustainable Development
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Objective – In the interest of helping libraries make evidence based decisions about open-source software (OSS), the objective of this research is to establish whether tools that automate the evaluation of OSS project communities could be used specifically on scholarly communications OSS (SC-OSS) projects to provide actionable insights for libraries to guide strategic decision making and corrective interventions. Methods – Seven OSS project communities were selected for evaluation, chosen from widely used scholarly communications software applications used in Canada for repositories, journal hosting, and archives. While all aspects of OSS projects may be evaluated at the project or network/ecosystem level, addressing the actors, software, or orchestration (Linåker et al., 2022), community evaluation that looks at the interaction patterns between project contributors is the practical focus of this research paper since there are multiple human factors that librarians who may not be software developers can impact. We identified a community analysis tool called csDetector (Almarimi et al., 2021) from the software engineering literature. This tool was chosen based on two main criteria: 1) ability to analyze data from GitHub repositories (the code sharing platform used by all selected SC-OSS projects) and 2) capacity to automatically produce results without manual intervention. Since some of the seven OSS projects were spread across multiple GitHub repositories, a total of 11 datasets from GitHub, each containing three months’ worth of data, were analyzed using csDetector. Results – The results produced by csDetector are interesting though not without limitations. The tool is complex and requires the user to have software development skills to use it effectively. It lacked sufficient documentation, which made interpreting the results challenging. The analysis from csDetector, which identifies community smells (i.e., types of organizational and social dysfunction within software projects [Tamburri et al, 2015, 2021a]), suggests that these SC-OSS project communities are experiencing knowledge sharing difficulties, weak collaboration practices, or other member interaction dysfunctions that can eventually permanently affect community health. Having a software tool that can take metrics from GitHub and detect community smells is a valuable way to illustrate problems in the project’s community and point the way to remedying dysfunction. Conclusion – While the OSS community analysis tool csDetector currently presents several hurdles before it can be used, and results generated come with caveats, it can be part of an approach to support evidence based decision-making pertaining to SC-OSS in libraries. The information provided can be worth monitoring (especially social network metrics such as centrality) and their results, particularly for community smells, identify problems that may be addressed by non-developers. Awareness of community smells in OSS can provide a deeper understanding of OSS sustainability as it provides a language to identify suboptimal social dynamics.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Direct model labels (unvalidated)
Per-model category and study-design labels from the labeling rounds. They are machine output, unvalidated, and the disagreement between models ships as data. No study design here is MEDLINE-validated yet.
| Model arm | Categories | Study design | Confidence |
|---|---|---|---|
| gemma | Scholarly communicationOpen science Domain: not available · Genre: Empirical About the Canadian research system: no · About a Canadian topic: no | Not applicable | low |
| gpt | Scholarly communication Domain: not available · Genre: Methods About the Canadian research system: no · About a Canadian topic: no | Theoretical or conceptual | high |
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.015 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.012 | 0.415 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it