This repository offers derivative reading copies of five papers by Kengo Tomita. The individual paper DOIs are the canonical scholarly citation targets. The repository makes the texts, their relationships, and their stated limits easier to inspect; it does not substitute a map for the papers or independently certify their claims.
Repository/corpus release: 3.5.0. T5 preprint version: 3. T4 English translation edition: 1.0.0. These are separate version numbers. All five original manuscript blocks and canonical DOIs are unchanged from release 3.2.1. This release incorporates the previously unuploaded English-access patch and source-grounded Japanese reading safeguards in one update. The translation is not a new paper or preprint version.
Status supplied by the author (September 2026): T1 is a preprint whose journal version is under peer review; T2–T5 are preprints. These labels are not new publication-status determinations.
Website: https://mekiki-framework.github.io/
For a narrow question, open SOURCE_INDEX.md, select a paper and section, and read that source passage. For the whole programme, use THEORY_MAP.md for orientation and then read T1 → T2 → T3 → T4 → T5. The reading order is not a claim that all later arguments are deduced from earlier papers.
For T5, the claim-status map distinguishes the analytic theorem, constructive normative extension, interpretive visibility thesis, sociological apparatus hypothesis, and historical evidence. Version 3 additionally specifies the attribution and derivative use of agent-relative, a limited moral-luck analogy, and the separation of technical terms from ordinary usages.
If a source and its guide disagree, report the discrepancy. A critical reader may reject a claim; the guides are there to make the claim and its scope inspectable, not to require agreement.
| Paper | Reading views | Canonical source |
|---|---|---|
| T1 — Domain-Native Development: A Mekiki Framework for AI-Assisted Knowledge Work | HTML · Markdown | https://doi.org/10.31235/osf.io/cwkav_v1 |
| T2 — Philosophy as Cognitive Assay: Measuring the Delegation Legitimacy Boundary in AI-Assisted Knowledge Work | HTML · Markdown | https://doi.org/10.31235/osf.io/e9qw5_v2 |
| T3 — Decomposing Agency, Isolating Answerability: Cultivating What Cannot Be Delegated in AI-Assisted Learning | HTML · Markdown | https://doi.org/10.35542/osf.io/hvbfe_v2 |
| T4 — 組織における「自分ごと化」のアドヒアランス的再記述：知識労働がAIで加速された時に組織に求められるもの | 日本語 HTML · 日本語 Markdown · English translation · English Markdown | https://doi.org/10.31235/osf.io/495wg_v1 |
| T5 — Why Play When AI Can Win: The Non-Transferability of Participation and the Visibility of Dignity | HTML · Markdown | https://doi.org/10.31235/osf.io/593ah_v3 |
All five papers in one Markdown file remains available. The individual paper files are generated extractions, not separately edited manuscripts. T4 contains Japanese full text and an English abstract. A separate English access translation with translator notes is now available. The Japanese original governs interpretation; translator notes are not part of the original paper. The supplied English abstract is retained verbatim at the end of the translation. Cite the original DOI for the paper’s claims and identify the translation edition when quoting its English wording or notes. Figure images omitted from the supplied corpus remain omitted; supplied legends and tables are preserved.
- llms.txt: concise, question-oriented retrieval entry point.
- SOURCE_INDEX.md: complete section navigation and topic routes.
- THEORY_MAP.md: cumulative interpretation, terms, and scope boundaries.
- T5_CLAIM_STATUS.md: source-linked T5 v3 claim entries.
- claims/t5.json: the single ledger used to generate the claim map and metadata's claim summaries.
- FOR_AI_READERS.md: core term definitions before the interpretation guide; optional application heuristics for deliverable, learning, inquiry, and play follow separately.
- metadata.json: project-specific JSON manifest of paper versions, sources, file roles, and conceptual safeguards.
- source_manifest.json: hashes, exact-copy derivation, and section locators.
- AI_READING_TESTS.md: 17 reading probes and source-bound rubrics, including three motivated by author-reported misreadings and two Japanese-word checks added in 3.5.0. No reproduced model runs or scores are claimed.
- CITATION.md: paper DOI citation and derivative-file attribution.
The theory map, claim ledger, and test rubrics share sources; agreement among them is not independent evidence. Consult the paper itself.
Use SOURCE_INDEX.md to find the paper and section relevant to my question. Read the source text before using the guides. Preserve the distinction between theorem, normative proposal, interpretation, hypothesis, and historical evidence. Cite the relevant paper DOI and section. Do not infer one paper's author–date suffixes from another paper. Report any source-guide conflict. Do not assume that a supplied guide validates the theory.
T5's literal Japanese translation title is 「AIが勝てるとき，なぜ遊ぶのか：参加の非移転性と尊厳の可視性」. The author also uses 「AIのほうが強いのに、人はなぜ遊ぶのか」 as an announcement title. Both identify the same English T5; the canonical English title, DOI and preprint version are unchanged. Only the main announcement title has been supplied, so no translated subtitle is attributed to the author.
The Japanese-reading table prevents reducing 知好楽の「楽」 to mood, 渋沢の「趣味」 to hobby alone, or play to a domain opposed to work. The short/long formula correspondence distinguishes a literal translation from the author's public explanatory wording. These are access aids, not additions to the paper or evidence for its claims.
- T1: https://researchmap.jp/ketomy/others/53500468
- T2: https://researchmap.jp/ketomy/others/53699246
- T3: https://researchmap.jp/ketomy/others/54141521
The files are ready to upload; no build is required for this release. Copy the contents of this directory into the repository root, retaining the papers, claims, tests, and tools subdirectories. Do not nest the release directory under the website root. .nojekyll keeps the static HTML/Markdown paths intact.
The 3.5.0 distribution is a changed-files-only update from the published 3.2.1 baseline. It already includes the T4 English patch: upload the new patch once, preserving subdirectories and leaving unlisted files in place. Do not upload the old English patch afterward, because its older shared files would overwrite the integrated guidance. No deletion or intermediate release is required.
For later maintenance, see MAINTENANCE.md. python tools/validate_corpus.py checks local consistency without network access. Rebuilding generated views uses python tools/build_access.py and requires Pandoc. These checks do not demonstrate better LLM comprehension or guarantee indexing, crawling, or DOI resolution.
The existing CC BY 4.0 license is unchanged. Reuse of derivative files requires attribution; scholarly claims should cite the paper DOI.
Kengo Tomita
Institute of Technology, Shimizu Corporation
ORCID: https://orcid.org/0000-0002-7310-8093
X: https://x.com/KennT1804