The knowledge base

Every file the untranslatable-words project wrote, rendered as HTML, exactly as it stood at the project’s close on 2026-09-08

This is the complete research record behind the essay Untranslatable, Allegedly: 345 pages. The project’s own wiki index is the best starting point — it catalogs every page with a summary — and the session log is the timeline. The lists below are generated from the directory layout.

Word case studies (25)

Concept pages

Topic pages

The untranslatable-words phenomenon (popular culture)

Wiki index — the catalog of every page, maintained by the project itself.
Session log — one entry per session, newest first.
Conventions — the wiki’s schema.
Project close-out — where the project ended.

Project rules and working files

MISSION.md — what this project is
The knowledge-base framework for this project
NEXT.md — current state and queue
resume-prompt.md — session entry point
BUDGET.md — paid API spending (Project-2)
CLAUDE.md — Project-2 (read every session)
tools/check_links.py — the link checker run before every commit (plain text).

Owner’s journal (62 entries)

Source records (240)

One file per source consulted, with exact quotations, access dates, stance, gaps, and the pages that use it.

About this mirror

These pages are verbatim HTML renderings of the markdown files in the project-2/ directory of the project repository, generated at close-out on 2026-09-08. Relative links between pages have been rewritten to point at the rendered files; nothing else has been changed. The knowledge base was researched and written entirely by Claude, an AI assistant made by Anthropic, working autonomously in scheduled sessions between 28 August and 8 September 2026; the project was conceived, funded, and monitored by Tom Gally. Appendix A of the essay explains how the knowledge base is organized. These pages are intended for direct human readers and carry metadata asking search engines not to index them and AI systems not to train on them.