Translation method
From source evidence to clear English
DGST is designed around an evidence-first translation workflow. The translation project is downstream of preserved Scripture identity, explicit versification mappings, source editions, linguistic analysis, textual evidence, translation review, and a separate publication action.
Original-language evidence is required
An English verse in the KJV, ASV, WEB, or another translation does not by itself create a DGST translation unit. DGST wording requires directly mapped Hebrew, Aramaic, or Greek evidence. Where that mapping is unresolved, the public research experience can identify the gap; it does not synthesize an English verse to hide it.
Source-only research drafting
DGSP can create a pre-publication research candidate from the source packet only after the source-language draft gate passes. An AI-assisted source-only draft may use the mapped Hebrew, Aramaic, or Greek passage together with stored morphology, sourced lexical evidence, syntax, and approved textual decisions. English comparison translations are excluded from generation. The resulting text is a research draft—not approval and not Scripture—and the Holy Bible website does not expose that wording before publication.
When a source-only research decision exists, the public Study page may disclose that workflow state without revealing the draft. This preserves useful transparency about how the pipeline is progressing while keeping machine or research wording outside the public Bible.
Verses are locators, not database identity
References such as “John 3:16” are human navigation labels. DGSP keeps stable content identity separately and maps reference systems onto that identity. This matters wherever Masoretic, English, or edition-specific numbering differs. A source edition’s own locator remains visible even when the website routes the passage through the DGC66/KJV-style public navigation reference.
English Bibles are comparison and approval-QC resources
Rights-cleared published translations can be compared during translation and quality control, but they are not treated as the original-language source from which DGST wording is generated. The website keeps English translations, Hebrew/Aramaic/Greek source editions, and linguistic resources visibly distinct.
The live DGST reference-match policy can accelerate the translation-review approval step after a draft has already passed the source-grounding and review-readiness gates. If the complete adopted rendering exactly matches an explicitly enabled English comparison edition for the same content unit, DGSP may auto-approve that decision.
- ASV — American Standard Version (1901)
- KJV — King James Version (1769 standardized text)
- WEB — World English Bible
Policy version: dgst-reference-match-v1. Merely importing another English edition does not grant it auto-approval authority.
The match is intentionally strict: DGSP trims leading and trailing whitespace and collapses whitespace runs only. Case, punctuation, spelling, word order, and wording remain significant. There is no fuzzy, semantic, substring, punctuation-insensitive, or AI-similarity approval. A human REOPEN action suppresses this automatic rule and returns the decision to human review.
This does not weaken the source rule. English comparison text remains QC/reference evidence only. It does not create the DGST unit, create the wording, satisfy source-language evidence requirements, resolve textual variants, or become primary translation evidence.
Approval is not publication
A decision may be research-draft, human-approved, or exact-reference-match auto-approved and still remain completely unpublished. Translation-review approval never creates a public DGST passage by itself. The public Bible therefore may disclose that a decision has completed approval while continuing to hide the approved wording until the separate revision publication process succeeds.
A sealed public revision controls DGST publication
Translation-unit, candidate, decision, review, and approval states are useful workflow evidence, but none of them alone makes a wording public Scripture. The Holy Bible website publishes DGST wording only when that exact rendering is stored in a sealed DGST revision whose publication metadata explicitly marks it public, whose publication date is effective, and whose stored publication gate snapshot records a passed authoritative DGST publication gate. Working-edition passages, scaffolds, machine drafts, source-only research drafts, other research drafts, reopened decisions, human-approved decisions, and auto-approved decisions remain outside the public Scripture text until that publication boundary is crossed.
“Why DGST translates it this way” follows the same release boundary. A public explanation must be the translation decision sealed into the public revision unit for that passage. The website does not expose a separate pre-release decision simply because its row has advanced farther in the internal workflow.
Research evidence keeps its uncertainty and its own rights
Morphology, apparatus parsing, witness support, lexical data, versification mappings, and other evidence keep their provenance and review state. They also keep independent rights and source-package eligibility: a linguistic or mapping resource is not assumed to inherit permission from the Scripture text it annotates. Public Study views expose a research layer only when the stored source rights permit public display/API distribution and DGSP considers that source package eligible; otherwise the layer remains internal.