Scope
The initial audit reviewed 17 works in the Scholarly Sources registry. Two August 2026 reviews expanded the registry to 24 records. The September 10 alignment review added Cattell (1948), Goldstone et al. (2008), Panzarasa and Jennings (2006), Proutskova (2026), and Riedl (2026), bringing the registry to 29 records. Six supplementary works retained for citation history are now visibly distinguished from sources used in current Reference explanations. Across these review rounds, author names, publication year, title, source title, volume, issue, pages or article number, editors and publisher where applicable, and DOI or other stable identifier were checked against authoritative records.
DOI-linked records were checked against DOI registration or publisher records; books and chapters were checked against publisher catalog or table-of-contents records. Contemporary additions were checked against OpenReview, Springer Nature, arXiv, Elsevier/Crossref-linked records, ACM proceedings records, and the King's College London research repository as appropriate to publication status.
Source-to-claim review
The audit also reviewed the principal places where each work is used in the CSI Reference. The test was whether the cited source directly supports the attached scholarly principle—for example, composition models, collective constructs, construct validity, dynamic emergence, situation awareness, human–AI interaction, probabilistic calibration, group-state modeling, tempo-relational team modeling, collective dynamics, Group Awareness Tools, or Group Conversational Agents—or whether the text clearly identifies the statement as a CSI synthesis or original proposition.
This is a targeted evidentiary audit rather than a new systematic literature review. It does not imply that every source endorses CSI as a framework, nor that a source supports claims beyond the scope described in the article text.
Audit outcomes
- No DOI mismatch was identified among the DOI-linked registry records reviewed.
- The Kozlowski and Klein chapter record was expanded to include the book's full subtitle.
- The Sawyer (2000) chapter record was expanded to include its verified page range.
- The August 2026 supplement established the close-neighbor set and the contemporary-literature alignment review added Prabhu et al. (2025) as an earlier direct group-level computational example.
- The new sources were used to narrow, rather than inflate, CSI priority claims: computational work already models group states, team constructs, collective dynamics, and group-oriented AI interaction.
- The September 2026 additions strengthen the group-measurement, emergence, multi-agent, and close-computational-comparison coverage using works cited by the current research program.
- The registry displays its current verification date and links to this audit record.
- Automated validation checks that DOI links use canonical
https://doi.org/form and that article citation anchors resolve to registry records.
How to interpret verification
Bibliographically verified means that the public record matches the available authoritative metadata or, for preprints and accepted/in-press work, the authoritative repository record available at the time of review. Claim-use reviewed means that the source is used for a scholarly principle within its apparent scope. Neither label means that the source validates CSI's original definitions, inferential chain, governance requirements, or proposed system architecture.
Ongoing maintenance
Future additions to the registry will undergo the same checks before being described as verified. Publication status for preprints and accepted/in-press work will be updated when a version of record becomes available. Corrections will be recorded in the Version History. Readers may report a bibliographic or interpretive concern through ed@collectivestateinference.org.
Audit method and correction handling are governed by the Editorial and Citation Policy.