Batch 4: Five Entries Source-Checked
Outcome: All five entries are source_checked for the declared material-claim traceability scope after revisions and independent post-edit review. No material unresolved findings remain. None is expert_reviewed.
Item-by-item changes and outcomes
- The Lucas–Penrose Argument — Independent review found no material issue. The claim ledger tracks the theorem’s stated conditions, Lucas’s argument and objections, selected critics, Penrose’s later proposal, and publication metadata. The article discloses that the original 1931 Gödel paper and relevant chapter of Penrose’s 1989 book were not directly consulted, and that the full quantum/brain chapter of Shadows of the Mind was not read.
- The Symbol Grounding Problem — Independent review found no material issue. Harnad’s 1990 article supports the dictionary example, proposed hybrid account and its limits; Taddeo and Floridi’s critique is attributed to them, not presented as consensus. The article distinguishes the 1990 journal publication from the 1999 arXiv deposit and labels current-AI examples as editorial prompts.
- The Frame Problem — The technical description now says “m fluents (properties of situations),” not “m facts.” Targeted follow-up confirmed McCarthy and Hayes’s n-actions/m-fluents statement at §4.3, printed p. 31, and the definition of fluents at §3.2, p. 19. The earlier precision finding is cleared.
- Blockhead — Independent review found no material issue. The entry’s account is traceable to Block’s readable posted full text, the SEP discussion, and Schwaninger’s named interpretation. It distinguishes logical possibility from feasibility and labels added examples as teaching extensions. The journal landing page did not expose the full text; review used a posted copy and checked bibliographic metadata separately.
- Paley’s Watch — Final targeted review cleared the page-mapping and OCR-description findings. The 1802 table of contents places Chapter IV at p. 36 and Chapter VI at p. 55; the eye passage begins under Chapter VI and continues on p. 56. The final article lists Chapter IV pp. 36–38 and Chapter VI pp. 55–56, distinguishes OCR Archive.org Paley/Darwin texts from the keyboarded/encoded University of Michigan Hume edition, and its five URLs match the dataset references.
Important limits
Source_checked records traceability of claims to relevant passages in appropriate sources. It does not establish truth, the success of an argument, scholarly consensus, exhaustive literature review, or expert approval. Primary texts establish what their authors argued, not that the arguments succeed; named secondary positions are not treated as consensus. The source-specific access limits and exact passage locations are recorded in the evidence ledger. Among the disclosed non-material access gaps: publisher full text was unavailable for some works; Lucas’s source check did not directly consult the original 1931 Gödel paper or the relevant 1989 Penrose chapter; only selected Penrose passages were read; and Paley/Darwin OCR and Hume transcription risks remain.
Audit count and local state
After this batch, 18 of 57 expansion entries have received claim-level audits; 39 remain outside claim-level audits. The evidence ledger remains outside data/experiments.json. Work is on local-only branch audit/source-review-high-risk; everything new remains uncommitted and unpushed, with no merge or public release.