Principles

These principles are the constitution of the Wheel of Heaven Translation Program. They override convenience, throughput, and stylistic preference. Every agent role, every workflow step, and every glossary entry is downstream of them.

1. Accuracy above lens#

Where the source language is unambiguous, the translation reflects it. Period.

This is the load-bearing commitment of the entire Program. The project does not translate freely to accommodate the Wheel of Heaven reading; it translates accurately and lets the WoH reading emerge from the apparatus where the source supports it.

The corollary: if a credentialed scholar of the source tradition reads a WoH Translation and says “that is not what the Hebrew / Greek / Akkadian says,” the translation has failed — regardless of how well the choice serves the lens. The whole credibility of the Program rests on this commitment being literally true.

Example. Genesis 1:1 — bereshit bara elohim et hashamayim v’et ha’aretz. The verb bara is the qal of a Hebrew root whose lexical core is “shape, fashion, bring into being.” Rendering it as “create” (KJV/ASV) or “shape” (WoH) are both defensible. The WoH choice “shape” is not a lens-driven freedom — it is a defensible reading of the lexical core, documented in the glossary with the rationale “preserves the verb’s distinctness from asah and yatsar without inserting a metaphysical claim either way.”

Counter-example. A reading like bereshit bara elohim → “in the beginning, the extraterrestrials engineered” would be a lens intrusion. The Hebrew word bara does not mean “engineered” and the Hebrew word elohim does not name a specific category of beings. That kind of move violates this principle and would not ship.

2. The lens lives in the apparatus, not in the text#

The translated text is what the reader sees first; the apparatus is where the project’s interpretive commitments live. The two are deliberately separated.

The text in paragraphs[].i18n.en reads as a defensible scholarly translation. A reader unfamiliar with Wheel of Heaven, reading the translation cold, should not detect “this is a WoH translation” from the English alone. They might detect it from the per-verse commentary, or from the methodology page, or from the glossary entries — but the translation itself reads straight.

This is how Édouard Dhorme (whose French Bible Raël used) did his work. Dhorme’s translation reads as a careful, scholarly French rendering of the Hebrew; the interpretive moves are in the footnotes and the introduction, not in the body of the text. Dhorme is still cited in serious biblical scholarship a century later for exactly this discipline.

Concrete: the per-paragraph commentary field can argue why the WoH choice is defensible, can name what scholarly literature supports it, can note what the lens makes visible. The paragraphs[].i18n.en field cannot do those things — it can only render.

3. Where the source is ambiguous, the Wheel of Heaven canon is the tiebreaker#

Hebrew, Greek, Akkadian, and most other source languages have words with multiple defensible English renderings. Lexicons list ranges, not points. Translation is therefore always making choices.

The project’s stated stance (decision #1 in the project’s strategy record): the Raëlian canon — The Book Which Tells The Truth (1974), Extra-Terrestrials Took Me To Their Planet (1975), and Let’s Welcome The Extra-Terrestrials (1979) — is the foundational lens. Where two defensible translations of a verse sit on the spectrum, the project prefers the one most consistent with that canon.

This is not “translating to fit the canon.” It is “using the canon to break ties where the source genuinely permits multiple readings, and recording the choice openly so the reader can audit it.”

Example. elohim is grammatically masculine plural in Hebrew (the -im suffix is the unambiguous plural marker). It can be construed singularly in some contexts (a “plural of majesty” reading), but the plural form is the grammatical default. Most modern English translations render it singular (“God”) on theological grounds inherited from the LXX. The WoH translation preserves the plural (“the Elohim”) because the canon reads elohim as plural agency. Both readings are defensible; the canon is the tiebreaker; the glossary entry records the choice as claim_type: direct (the plural form is grammatically explicit).

Example. ruach elohim in Gen 1:2 admits three lexical fields: breath, wind, spirit. KJV/ASV “Spirit of God” is a Trinitarian-tradition retrojection; the Hebrew supports any of the three. The WoH translation chooses “breath of the Elohim” because (a) the immediately-following verb merachefet “hovering” fits a physical reading better, and (b) the canon does not need a Christian Trinitarian metaphysics in the second verse of Genesis. Glossary entry records this as claim_type: inferred — defensible from the lexical range, not consensus.

4. Claim-type discipline#

Every WoH-divergent lexical choice carries a claim_type label, one of:

• direct — the WoH choice is grammatically explicit in the source or matches a scholarly consensus. There is no real disagreement among careful source-language readers.
• inferred — the WoH choice is a defensible reading of the source consistent with what some named scholars argue, but not the consensus or the traditional rendering. The lexical evidence supports it; the choice between this and the standard reading is interpretive.
• speculative — the WoH choice goes beyond what any single scholarly source attests. It is the project’s interpretive synthesis or a project-specific reading. Honest about being speculative; not dressed up as direct or inferred.

This labeling is not for the project’s internal use only — it is the reader’s audit handle. A reader inspecting the glossary can filter to speculative entries to see exactly where the project made non-scholarly moves, and can decide for themselves whether to trust those moves.

The same claim_type discipline is used throughout the Wheel of Heaven content corpus (see Editorial Passes); the translation program inherits the discipline at the lexical level.

claim_type: speculative entries require explicit human sign-off. They cannot ship on an AI editor or reviewer’s say-so alone.

5. Audit-trail integrity#

Every step of the pipeline is preserved. Nothing is silently overwritten or edited away.

• The source file (source-{lang}-{N}.json) is immutable. The pointed/transliterated source is what it is.
• The chapter file (chapter-{N}.json) preserves translation.modelDrafts[], translation.reviewer, translation.reviewedAt, translation.reviewerReport, the glossary version pinned to the chapter, and (eventually) a history of version bumps.
• The glossary file is semver-versioned, and every change to an existing entry bumps the major version. Chapters pin to the glossary version that produced them.
• The editor’s escalation report and the reviewer’s verdict report are preserved alongside the chapter. They are not summary artifacts that get discarded; they are part of the chapter’s permanent record.

The audit trail is what lets the project make a credible claim of scholarly seriousness. A reader who challenges a specific WoH-divergence in a chapter can trace it through:

1. The paragraphs[].i18n.en rendering of the verse.
2. The paragraphs[].glossaryRefs[] array, naming the glossary entry that drove the choice.
3. The glossary entry itself, with wohChoice, claim_type, rationale, and appliesTo[].
4. The chapter’s translation.reviewerReport to see what the Reviewer agent verdicted on that specific verse and what citations were given.
5. The escalation report, where applicable, to see the editor’s reasoning for the more contested choices.

Each layer of the audit chain can be checked independently. This is the project’s standing offer to serious readers: read the translation, but if you want to know why any specific choice was made, the answer is recorded, sourced, and signed.

How these principles relate#

The principles are ordered by priority. Accuracy is the absolute top; everything else operates within it. The lens lives in the apparatus because accuracy comes first in the text. The canon is the tiebreaker only where the source is ambiguous — ambiguity is the entry condition for the lens to operate. The claim-type discipline and the audit trail are what make the lens-as-tiebreaker honest rather than a fig leaf.

A simple test for any proposed translation choice: can it survive all five principles simultaneously? If not, it does not ship.