Production date: 30 October 2026. Expanded: 31 October 2026 (57 new entries; see research/expansion.md).
| Category | Entries |
|---|---|
| artificial-intelligence | 8 |
| consciousness | 13 |
| economics | 8 |
| epistemology | 17 |
| ethics | 14 |
| identity | 11 |
| logic | 13 |
| mathematics | 9 |
| physics | 14 |
| science-and-biology | 5 |
| society-and-justice | 9 |
| time-and-causality | 11 |
Type labels: 3 analogies, 19 arguments, 7 models, 19 paradoxes, 15 puzzles, 69 thought experiments. Selected puzzles, arguments, paradoxes, and models are deliberately included and labelled; the total is not a claim that every entry is a thought experiment in the narrow sense.
Difficulty: 35 Beginner, 67 Intermediate, 30 Advanced. The science-and-biology and artificial-intelligence categories remain the smallest and are the most useful places for further contributions.
scripts/validate.py. That evaluator supports the schema keywords used here and fails closed on unsupported validation keywords; it is not a general-purpose JSON Schema implementation.Source consultation is recorded in research/philosophy.md, research/knowledge.md, research/science.md, and, for the 31 October 2026 expansion, research/expansion.md. The expansion record remains an accurate account of the initial 31 October checks: page access, titles, and topic-keyword counts were recorded, but those checks did not establish claim support. The 1 November claim-level audits are documented in 17 small batches: batch 1 report (ledger); batch 2 report (ledger); batch 3 report (ledger); batch 4 report (ledger); batch 5 report (ledger); batch 6 report (ledger); batch 7 report (ledger); batch 8 report (ledger); batch 9 report (ledger); batch 10 report (ledger); batch 11 report (ledger); batch 12 report (ledger); batch 13 report (ledger); batch 14 report (ledger); batch 15 report (ledger); batch 16 report (ledger); batch 17 report (ledger). The consolidated evidence ledger contains all repository audit records and the final inventory reconciliation.
Across the 17 ledgers, there are 60 unique audit records. Comparing their IDs with the 57-entry expansion inventory gives 57 of 57 expansion entries with audit records; none remain without a claim-level record. Three additional audited entries—Utility Monster, the AI Box Scenario, and the Spandrels of San Marco—are outside that inventory. The expansion entries have 55 source_checked and two partially_source_checked statuses: Nietzsche’s Eternal Recurrence and Plato’s Allegory of the Cave. Across all 60 records, 58 are source_checked and two are partial; none is expert_reviewed. The earlier batch-nine count of 35 included three out-of-inventory records and did not mean 35 of 57 expansion entries.
Source_checked is an editorial traceability status, not a claim that a philosophical conclusion is true or settled, every primary work or proof was read in full, the literature was exhaustively reviewed, or an entry received human expert review. The audit ledger remains outside the public data schema; no public status badge or expert-review claim was added. This is coverage for the 57-entry expansion inventory only; the planned risk-ranked audit of the original 75 entries and the broader improvement-plan phases remain separate work.
Many original historical books and articles were accessed through scholarly encyclopedia discussions or teaching materials rather than inspected in full; the batch ledgers give exact passage locations and limits. The Utility Monster audit read Nozick’s chapter 3, page 41, in a university-hosted PDF reproduction, while Chappell’s related paper remained abstract/record-only. The AI-box audit consulted Yudkowsky’s retrospective account and contemporary 2002 mailing-list messages, but not private transcripts. The Spandrels audit read relevant passages in a reproduction of the full 1979 paper, Gould’s 1997 clarification, and Gould and Vrba’s 1982 exaptation paper; the third-party reproduction and OCR may contain errors. Wigner’s 1961/1962 edition dating remains uncertain. The Four-Case audit did not directly access Pereboom’s 2001 chapter; the Kripke audit used a selected excerpt rather than the complete book. The completed philosophy audit records stronger direct access obtained for Singer, Nagel, and Leibniz materials. In batch 10, Quine’s 1970 passages were read in a third-party course-hosted scan; the DOI landing page at the Philosophy Documentation Center provides metadata but gates full text. Einstein’s train exposition was consulted in Lawson’s 1920 English translation; the 1916 German first-edition title page was not checked. The Nietzsche notebook fragment cited for a cosmological reading was not directly accessed, so Eternal Recurrence remains partially source-checked and relies on the cited secondary account for that fragment. Batch 11 read Galileo’s relevant ship and tower passages in the full Drake translation, but did not collate the original Italian; Heraclitus’s book is lost, so this audit used later ancient witnesses and reports the Graham–Taran disagreement rather than resolving it. Thomson’s and Benacerraf’s relevant passages were read in hosted copies; the complete papers were not read systematically. Batch 12 added Philo’s later testimony and Bowin’s discussion for Dion and Theon, but did not access Chrysippus’s lost original; the Grelling–Nelson paper was represented by a translated passage in the SEP, not read in a primary copy, and the exact 1907/1908 publication chronology remains unsettled; McTaggart’s relevant pages were read in a full-text scan reproduction, while the publisher’s full text remained gated.
The new batches 13–17 likewise retain access and interpretation limits. Zeno’s original work is lost and the Arrow/Dichotomy wording is reconstructed from later testimony. The Cave article’s approximate c. 375 BCE date appears in an automatically generated Project Gutenberg catalogue summary and is not confirmed by the consulted scholarly chronology sources, so that record remains partial. The Idle Argument relies on later ancient witnesses; Galileo’s Paradox was read in an English translation rather than the Italian original; Moore’s relevant passages were checked in a 1993 reprint rather than the 1939 issue. The Statue and Clay and Problem of the Many entries map competing metaphysical accounts rather than resolving them; Condorcet’s voting-cycle claims remain conditional on the stated preference profile. The formal epistemology entries do not establish a consensus response. For the final batch, Mill’s Other Minds passage and Dretske’s original mule article were accessed through secondary quotations/discussion rather than the primary full text; Lewis’s mad-pain cases are stipulated, and an objection remains unresolved; the Stag Hunt payoff matrix is illustrative and the original Rousseau text was read in English translation. No named human subject-matter expert reviewed batches 13–17.
No automated live-link sweep was run: retrieval establishes only the access levels documented during production, and the validator checks URL syntax rather than current availability. No claim is made that every historical priority or controversial interpretation was independently settled. Unresolved philosophical questions remain unresolved. AI-box roleplay, evolutionary analogies, formal paradoxes, and games are not represented as established empirical demonstrations.
The requested repository structure, dedicated articles, categorized README, original SVG banner, seven learning paths, connections guide, JSON/schema, contribution and attribution guidance, license, issue/PR templates, workflow, upload instructions, and downloadable ZIP are included. The original release reached the lower end of the requested 75–100 range; the expansion takes the collection to 132 entries. Footbridge is a trolley variation; original position and veil of ignorance are treated together; Chinese Room and Wigner’s Friend are cross-referenced rather than duplicated.
The user must create/upload the GitHub repository to execute the remote workflow. Independent expert review, exhaustive primary-text corroboration, and a live status check of every external URL remain outside this completed production pass. The owner may add a private community-reporting contact before inviting a large community. No backend, API key, paid service, or GitHub Pages deployment is required.