Source-Text Translation (WoH Translation Program)

How Wheel of Heaven produces original translations of religious and mythological source texts (Hebrew, Greek, Akkadian, Sumerian, Arabic, …) under the WoH lens. Principles, pipeline, agent roles, glossary system, source acquisition, file formats, citation discipline, and roadmap.

8 pages in this section

The Wheel of Heaven Translation Program produces original English-language translations of religious and mythological texts from their source languages, read through the Wheel of Heaven lens — which takes the Raëlian canon as foundational and prefers, where the source admits multiple defensible readings, the reading most consistent with that canon.

This is a different kind of work from the i18n localisation pipeline that translates English Wheel of Heaven content into the nine other site languages. The two pipelines do not compete; they sit at different points in the content flow.

PipelineDirectionDocumented at
WoH Translation ProgramSource language (Hebrew, Greek, …) → English, with WoH lens applied lexicallyThis section
i18n localisationEnglish (or French, for canon books) → 8 other site languages, preserving project terminologyTranslations

A book translated under the WoH Translation Program (worked example: Genesis-WoH, 50 chapters signed off as stable) is subsequently fanned out to the other site languages through the i18n localisation pipeline. The two are sequential, not parallel.

What lives here#

  • Principles — the project’s editorial commitments. Accuracy above lens; lens in the apparatus, not in the text; the Raëlian canon as the tiebreaker where the source is ambiguous; explicit claim_type discipline.
  • Workflow — the eight-step per-chapter pipeline from source acquisition through human sign-off, with a worked example.
  • Roles — full role definitions for the three AI agents in the pipeline (Translator, Editor, Reviewer), with their hard rules and input/output contracts.
  • Glossary system — the three glossary layers (production / overlay / distribution), what each is for, and how they interact during translation.
  • Source acquisition — where to obtain openly-licensed pointed/transliterated source texts, by tradition: Hebrew Bible, Greek NT, Akkadian, Sumerian, Egyptian, Qur’an, Pseudepigrapha.
  • File formats — the JSON schemas for source files, chapter files, and glossary files, with full worked examples.
  • Reviewer citation shelf — the canonical scholarly lexicons and commentaries the Reviewer agent must cite. Acceptable and unacceptable sources by tradition.
  • Roadmap — the prioritised text list, current state, and what is next.
  • Library Acquisition Program — the broader strategy the roadmap operates within: which texts are sourced, in which tradition, on which track (Edition vs Translation), and why.

Who this section is for#

  • Editors and reviewers who want to understand or audit the project’s translation methodology.
  • Scholarly readers who want to see the project’s source-handling discipline before deciding whether to trust the translations.
  • Future contributors (human or AI-assisted) producing translations under the Program.
  • Anyone curious about how the Raëlian-lens reading interacts with serious source-language work.

If you are reading a WoH Translation in the Library and want to know why the translation makes a particular choice, the glossary entry referenced from the chapter’s glossaryRefs[] is the immediate answer; this section is the methodological background behind that answer.

The one-paragraph version#

A chapter of source-language text flows through three AI agents in sequence — Translator (literal rendering applying the existing glossary), Editor (lexical-decision adjudication, commentary, glossary deltas), Reviewer (independent verification with mandatory scholarly citations). The Reviewer then produces a sign-off package; a named human reviewer (currently Zara Zinsfuss) signs off; the chapter is promoted to stable. The translated text reads as a defensible scholarly translation; the Wheel of Heaven lens lives in the per-verse commentary and the glossary entries, never in the translated English itself. Every divergence from a major public-domain translation is recorded with a claim_type (direct / inferred / speculative) so readers can audit exactly how far the lens went in any specific choice.

Current state (2026-06)#

  • Books shipped or in flight: 21. Hebrew Bible — Genesis (50 ch, complete), Exodus, Isaiah (scaffold), Ezekiel, Daniel, Job. New Testament — Matthew, Mark, Luke, Acts, Revelation. Qur’an (selections). Pseudepigrapha — 1 Enoch, Jubilees, Shi’ur Qomah. Mesopotamian — Gilgamesh XI, Adapa, Eridu Genesis, Enki and Ninmah, Enki and Ninhursag, Song of the Hoe. Ugaritic — Baal Cycle (KTU 1.3–1.4). All drafted chapters are signed off as stable.
  • Acquisition queue: 72 more books are registered as planned in catalog.json per the 2026-06 Library Acquisition Program: the Atrahasis ring, the Anatolian Kumarbi cycle, the Avesta, Greek (Theogony, Plato, the Hellenistic witnesses), Egyptian (Heavenly Cow, Pyramid Texts), Bahá’í, Caodai, Oomoto, more pseudepigrapha, Hekhalot, Gnostic/Coptic.
  • Production glossary is at v2.69.0 with 789 terms across six source languages (Hebrew, Sumerian, Akkadian, Ugaritic, Arabic, Greek). Distribution: 708 direct / 74 inferred / 7 speculative.
  • Per-translation overlay glossaries are active for 19 books (every shipped translation except Genesis-WoH and the Mormon Edition track). The mechanism is described in Glossary system.
  • Multi-language fanout is operational and runs after a book reaches stable — see Translations.
  • Principles The five editorial commitments that shape every Wheel of Heaven Translation: accuracy above lens, lens in the apparatus, the canon as tiebreaker, claim-type discipline, and audit-trail integrity.
  • The eight-step workflow The Wheel of Heaven Translation Program per-chapter pipeline from source acquisition through human sign-off to multi-language fanout. Each step's inputs, outputs, and acceptance criteria.
  • Agent roles Full role definitions for the three AI agents that produce a Wheel of Heaven Translation: Translator, Editor, Reviewer. Hard rules, input/output contracts, and failure modes for each.
  • The glossary system Three glossary layers (production, per-translation overlay, distribution) and how they interact to produce a Wheel of Heaven Translation and fan it out across the nine site languages.
  • Source acquisition Where to obtain openly-licensed pointed/transliterated source texts for Wheel of Heaven Translations, by tradition: Hebrew Bible, Greek NT, Akkadian/Sumerian, Egyptian, Qur'an, Pseudepigrapha. Plus the file format the project standardises on.
  • File formats The JSON schemas for every artifact in the Wheel of Heaven Translation pipeline: source files, chapter files, production glossary, per-translation overlay, distribution glossary, escalation reports, sign-off packages. Full worked examples from Genesis-WoH.
  • Reviewer citation shelf The canonical scholarly lexicons and commentaries the Reviewer agent must cite for every disputed verse, organised by source tradition. Plus the explicitly unacceptable citation sources.
  • Roadmap What the Wheel of Heaven Translation Program is translating, in what order, and why. Current production state (2026-08) and the acquisition queue.

Edit this page on GitHub