the People's Open Bible

Maintained by peoplesbible.com, which stewards the public revision process, hosting infrastructure, and this open-source project.

This is the people’s Bible — the most accurate-to-original translation from Hebrew, Greek, and Latin into English that we can build. Every decision is publicly defensible and publicly revisable. Our data, methods, and code are published as openly as possible.

Read the Bible · About the project · Our philosophy · Simplified POB

What makes this different

From manuscript originals

• Direct from the originals. Drafted from original Hebrew, Greek, and Latin source texts — including books where we used AI vision models to read the text directly from manuscript photographs. See the source corpora.
• Every translation decision is publicly documented. Every verse is a file in this repository with the source text, the English rendering, every lexical decision (source word, chosen gloss, alternatives considered, lexicon entry, rationale), and contested theological readings with alternatives preserved. Every change is traceable in git, and every disagreement can have a permanent URL. See John 1:1’s source record.
• Every verse is inspectable. Source text, draft and revision metadata, and decisions are published for independent review. A prompt hash and model identifier do not guarantee that a hosted AI draft can be reproduced byte-for-byte; see the present verification limits in the methodology.

Why we built this

A method, not a claim of authority

We’re not claiming authority. We’re presenting a method: use state-of-the-art AI alongside original Hebrew, Greek, and Latin source texts to render a translation as close to what the original scribes wrote as we can get. Everything that goes into it is open-sourced — source texts, restored public-domain manuscript material, model passes, lexical decisions, revisions, and disagreements.

The point is not that the first draft is perfect. The point is that the whole process is inspectable, repeatable, and publicly revisable. If a reader disagrees with a rendering, they can see why it was chosen, inspect the alternatives, and propose a better one.

See PHILOSOPHY.md for the full statement on openness, transparency, and our theological commitments.

For the ongoing biblical source-comparison and restoration program, see the research log: method, evidence, decisions, corrections, unsuccessful attempts, and work still outstanding.

Use it freely

Free to use in any way

Released under CC-BY 4.0: read it, fork it, quote it, adapt it, use it publicly, include it in commercial products, or build derivatives with attribution. We never paywall scripture. All of it is packaged and organized for anyone to use.

See LICENSE and DEDICATION.md.

How POB reads — compared to NKJV and NIV

Two verses that illustrate what POB does differently, and why.

Ecclesiastes 1:2

• The key word, hevel (הֶבֶל) — "breath" / "mere breath." Not simply "vanity" or "meaninglessness." The Hebrew image is concrete: breath, vapor, something real but fleeting, ungraspable, and gone almost as soon as it appears.
• Why not "vanity"? It is historically familiar, but modern readers often hear pride or self-obsession, which is not Qoheleth's point.
• Why not "meaninglessness"? It over-decides the theology and pushes the verse toward nihilism. Hevel is broader: transient, elusive, ironic, and frustrating.
• The speaker, Qoheleth (קֹהֶלֶת) — "Qoheleth." This preserves the Hebrew title/name instead of narrowing it immediately to "Preacher" or "Teacher."

Genesis 4:1

• The first verb, yada (יָדַע) — "knew." In context it functions as a euphemism for sexual relations—the conception that follows makes that clear—but the Hebrew image is still "knowing." POB preserves that image instead of replacing it with an explanation.
• Why not "made love"? The NIV makes the sexual meaning explicit, but the modern English phrase also suggests an emotional or romantic nuance that the Hebrew verb does not itself state. "Knew" lets the narrative supply the sexual meaning without adding that extra shading.
• Eve's word, qaniti (קָנִיתִי) — "I have acquired." This keeps the lexical link to Cain's name. Renderings such as "brought forth" explain the birth but weaken the Hebrew wordplay.
• The divine name (יהוה) — "Yahweh." This makes the name visible rather than hiding it behind the typographic substitute "LORD."

The philosophy behind these choices

POB's choice is source-image accuracy in the main text + contextual understanding in a companion layer — not traditional smoothing, theological over-resolution, or pastoral softening of the translation itself.

Ecclesiastes 1:2 shows why this matters: hevel is not just "vanity" and not simply "meaninglessness." It is breath — real, visible for a moment, impossible to hold. Genesis 4:1 shows the same principle in narrative: the Hebrew says Adam "knew" Eve, while the context communicates the sexual meaning. The translation preserves the source-text image rather than replacing it with a more explicit modern phrase.

In both cases, familiar renderings preserve something valuable, but they also make hidden choices for the reader. POB keeps the core source-language pressure in the sentence itself, then uses footnotes, YAML rationale, the in-app Original context AI tool, and public discussion to make every choice visible and defensible.

License

The translation is released under CC-BY 4.0 — the canonical open-content license. You may use, adapt, redistribute, or commercialize this work with attribution. Anyone may fork it, translate it into other languages, include it in commercial products, or build derivatives without permission. We never paywall scripture. See LICENSE and PHILOSOPHY.md.

Source texts retain their original licenses (see sources/README.md).

Doctrinal stance

Translation decisions follow the commitments in DOCTRINE.md. Declaring our stance up front is a form of honesty — critics can assess our output against our stated commitments rather than guessing at hidden biases.

Methodology

See METHODOLOGY.md for the drafting pipeline, cross-check protocol, and current verification limits. For the deuterocanonical source-text rescue work specifically — how the Swete corpus was improved, rescued, and confidence-promoted from direct scan inspection — see docs/PHASE8_CORPUS_QUALITY_RESCUE.md.

Dead Sea Scrolls and ancient witnesses

The Dead Sea Scrolls and Ancient Witness Project extends POB's source-image method to lawful Qumran/Judaean Desert images, fresh diplomatic transcription, explicitly marked restoration hypotheses, and Hebrew–Aramaic–Greek comparison. The accompanying backlog also covers the Samaritan Pentateuch, early Hebrew inscriptions and papyri, the Cairo Genizah, Masoretic codices, Septuagint witnesses, Targums, and the Peshitta. It is a textual evidence program, not an algorithmic declaration of canon or divine authority. The Old and New Testament restoration priority plan records the Codex-only target order, the comparison-source packages to create, and the strict separation between textual evidence and optional ImageGen visual reconstructions.

The Deuterocanonical Books (Apocrypha)

The People's Open Bible includes the Deuterocanonical books — Tobit, Judith, Wisdom of Solomon, Sirach (Ecclesiasticus), Baruch, 1–4 Maccabees, and others — as a clearly labeled section. See DEUTEROCANONICAL.md for the full strategy; in brief:

Why we include them

These books are canonical in the Roman Catholic (Trent, 1546), Eastern Orthodox (Synod of Jerusalem, 1672), and Oriental Orthodox traditions, and were included in Luther's 1534 German Bible and the original 1611 King James Version as a middle section between the Testaments. Their removal from most modern Protestant Bibles is a 19th-century development driven primarily by printing economics at the British and Foreign Bible Society — not theological consensus.

Excluding them would be making a theological editorial decision before the reader ever sees the text. That is the opposite of what the Cartha Open Bible exists to do. Our mission is transparency and auditability — present the textual tradition; let the reader and their faith community judge what they affirm as canonical.

We take no position on canonicity. Including these books is not an assertion that they carry the same authority as the Protestant canon. Excluding them would be an assertion that they do not. We translate and publish; the reader decides.

How we honor them

These books receive the same rigor as the Protestant canon — same drafter/reviser pipeline, same per-verse YAML with full provenance, same license (CC-BY 4.0). No corner-cutting because they are "deutero."

Each book is translated from the most-original extant text we can lawfully access:

• For books composed in Greek (Wisdom of Solomon, 2 Maccabees, the additions to Esther and Daniel), the LXX is the original — there is nothing behind it. Source: Swete's public-domain 1909–1930 Septuagint.
• For books where the Hebrew or Aramaic original is lost (1 Maccabees, Judith, Baruch, 1 Esdras), the LXX is the oldest surviving witness. Same source.
• For Sirach, we translate primarily from the Hebrew where it survives — from Schechter & Taylor's public-domain 1899 edition of Cairo Genizah MSS A and B, and from fresh AI-vision transcriptions of the other manuscripts from public-domain photographs.
• For Tobit, we translate from the LXX Greek (Long Recension, supported by Codex Sinaiticus) and plan to incorporate the Qumran Aramaic fragments as a primary source for the ~20% of verses they cover, pending a licensing request to the Israel Antiquities Authority.

How they are labeled

In both the repository and the mobile app:

• A clear section header: "Deuterocanonical Books (Apocrypha)"
• An introductory note explaining the canonical-status landscape across traditions.
• Each book's header shows its canonical status at a glance: "Canonical in: Roman Catholic, Eastern Orthodox. Considered useful but not canonical in: most Protestant traditions."
• Internal references (cross-references, reading plans, search results) treat these books as first-class citizens while preserving the label.

The reader should never be confused about what they are reading or which traditions receive it as Scripture.

The scholarship-integrity commitment

We consult the leading scholarly editions actively during our work — Beentjes 1997 for Hebrew Sirach, Fitzmyer 1995 for Qumran Tobit, Skehan & Di Lella 1987, the Göttingen critical LXX editions, and further scholarship as needed. That consultation is how modern translation has always been done: read everything, weigh every variant, produce fresh work. Copyright law governs reproduction, not reading — and consultation of scholarship to inform our own fresh translation is standard, uncontroversial scholarly practice.

What we do not do is vendor or reproduce copyrighted scholarly transcriptions (Beentjes's specific Hebrew text, the DJD volumes' Qumran transcriptions, Rahlfs digital editions under restrictive licenses, etc.). Every source text in this repository is either public-domain by age or freshly produced by us from public-domain photographs under CC-BY 4.0.

Where established scholarship informs a translation decision, we cite it in footnotes — fact-level citation, not reproduction of creative expression, uncontroversial under US copyright law (Feist v. Rural, 1991). This combination — fresh transcription + active consultation of the scholarly literature + transparent citation — is exactly how the NRSV, NABRE, and Orthodox Study Bible produce their Apocrypha sections. It's how serious translation actually works.

What we bring to this corner of the canon

English translations of the Apocrypha already exist — NRSV, NABRE, Orthodox Study Bible, NETS, the Lexham English Septuagint. Each was produced by a specialist committee with years of scholarly work. We are not trying to replace them. We are contributing something adjacent that has not existed before:

• The first freely-redistributable deuterocanonical translation from a scan-grounded corpus. Brenton's 1851 LXX translation is public-domain but 175 years old; every scholarly translation since is under a restrictive license. The People's Open Bible Apocrypha is released under CC-BY 4.0 — use it, quote it, remix it, ship it in your app. This is a substantive contribution on its own.

• A translation grounded in Swete's diplomatic Codex Vaticanus. NETS uses Rahlfs-Hanhart (eclectic). NRSV and NABRE use the Göttingen critical editions (also eclectic). No current English translation takes Swete's diplomatic Vaticanus as its primary source. For readers who want "what Codex Vaticanus actually prints" rather than "what modern editors reconstruct," this fills a real gap — and we are transparent that this is the choice we made and why.

• Per-verse provenance, publicly auditable. Every verse traces back to the specific scan pages it was drawn from, which independent transcriptions agreed or disagreed, which Hebrew or Aramaic parallels were consulted, which scholarly references informed the judgment, and at what confidence level the source-text reading was adjudicated. This level of auditability per verse is, to our knowledge, unprecedented for an English Bible translation. You don't have to trust our committee; you can read the receipts.

• Multi-witness triangulation at scale. For every deuterocanonical verse we cross-checked four independent Greek transcriptions (our own OCR, First1KGreek TEI, Rahlfs, and Amicarelli) against the actual scanned page image — and where Hebrew witnesses survive, we add them as additional evidence (Cairo Geniza Hebrew Sirach, the Masada scroll, Neubauer Tobit, the WLC Hebrew for 1 Esdras parallels). This kind of systematic multi-source grounding, applied per verse with a documented rubric, is harder for human committees to do at scale.

From the engineering and textual-criticism side, this will be the most transparent, the most auditable, the most permissively-licensed, and (for Sirach specifically) among the most Hebrew-grounded Apocrypha translations available in English today. That's the contribution, stated plainly.

For the full scholarly-source policy (three-zone model: vendored / consulted / forbidden), see REFERENCE_SOURCES.md.

Why this honors the texts

These books are read as Scripture by hundreds of millions of Christians, and have been for two millennia. They shaped the vocabulary of the New Testament — James echoes Sirach; Hebrews 11 lists Maccabean martyrs alongside the patriarchs; the Magnificat echoes Judith's prayer. They were the devotional library of Jesus's own world.

Translating them carefully, with provenance and honest footnotes, under a free license, using the best tools we have — this honors them as Scripture for those who receive them as Scripture, and as early Jewish religious literature for those who do not. It refuses the shortcut of deciding for the reader what counts.

That refusal is the whole point of the People's Open Bible.

Extra-canonical scripture

Beyond the 13 LXX-based deuterocanonical books (Phase 8), the Cartha Open Bible is committed to translating a further corpus of extra-canonical texts under CC-BY 4.0. See EXTRA_CANONICAL.md for the complete scope document.

The rationale is the same as for the deuterocanon, extended:

• Some of these texts are canonical in other Christian traditions. 1 Enoch and Jubilees are received as Scripture by the Ethiopian and Eritrean Orthodox Tewahedo Churches. A Bible that silently excludes them is a Bible shaped by Western publishing history, not by the historical breadth of Christian reception.
• Some are quoted explicitly in the New Testament. Jude 14-15 quotes 1 Enoch 1:9 directly.
• Some shaped how the New Testament canon formed. The Didache, 1 Clement, and the Shepherd of Hermas were included in early NT codices (Sinaiticus includes Hermas; Alexandrinus includes 1 Clement). Understanding canon formation requires access to the works that were weighed alongside the canonical ones.
• Every existing modern English translation is under restrictive copyright. No current open-license, modern-register translation of 1 Enoch, 2 Esdras, 2 Baruch, or the Didache exists.

Scope

Three tiers, by reception status rather than by scholarly quality. Every text ships with explicit canonical-status labeling across traditions, and for the third tier, with explicit framing of its date, manuscript provenance, and theological-literary character.

Tier 1 — extra-canonical with historical canonical status somewhere: 2 Esdras (KJV Apocrypha; Vulgate appendix), 1 Enoch (Ethiopian Orthodox canonical), Jubilees (Ethiopian Orthodox canonical), Psalms of Solomon (in some early LXX manuscripts).

Tier 2 — Apostolic Fathers, ancient church orders, and pseudepigrapha: Didache, 1 Clement, Shepherd of Hermas, Didascalia Apostolorum, Apostolic Tradition, Apostolic Church Order, Apostolic Constitutions, 2 Baruch, and Testaments of the Twelve Patriarchs.

Tier 3 — early Christian mystical, apocryphal, and Nag Hammadi texts (with framing): Gospel of Thomas, Gospel of Truth, Thunder, Perfect Mind, Gospel of Philip, Gospel of Mary, and the prioritized expansion catalog under sources/early_christian_texts/catalog.json. Theological difference is not a reason to hide a historical text. Instead, every work is labeled by manuscript, approximate date, literary form, and theological tradition, and is explicitly separated from canonical Scripture.

The active expansion adds 27 historically important works: eleven Nag Hammadi works, five other early Christian apocrypha, and seven Apostolic Fathers works, plus four ancient church orders. The catalog is the authoritative list and publication gate. A catalog entry can be prepared, drafted, reviewed, and tested without being presented as canonical Scripture or as a finished direct translation.

The expansion uses only identified public-domain witnesses for its initial bridge drafts: explicitly public-domain modern English witnesses for many Coptic and apocryphal works, and public-domain historic editions/translations for the Apostolic Fathers. These drafts are adaptation-and-translation bridges, not substitutes for the lost Greek originals or for direct review of the surviving Coptic/Greek. Every work remains visibly provisional until its source-language review gate is completed. Lacunae, uncertain restorations, and supplied words must remain distinguishable rather than being silently smoothed into the text. See EXTRA_CANONICAL.md for the complete list and conventions.

How this gets built

Extra-canonical translation follows Phase 9 (LXX deuterocanon drafting) in a phased rollout by shared pipeline. 2 Esdras and 2 Baruch now sit in the same Semitic apocalypse multi-witness family (see 2ESDRAS.md and 2BARUCH.md): 2 Esdras is Latin-primary, 2 Baruch is Syriac-primary, but the architecture is intentionally shared. The Apostolic Fathers (Didache, 1 Clement, Shepherd of Hermas) reuse our existing LXX Greek-primary infrastructure. 1 Enoch and Jubilees require a new Ethiopic (Ge'ez) pipeline. The Nag Hammadi group requires a Coptic review pipeline plus per-codex license research. Public-domain English bridges allow responsible early reader access, but do not remove that source-language gate.

See EXTRA_CANONICAL.md for per-text source editions, phase sequencing, and the per-text license situation.

Current status

The project is in its initial AI-drafting phase. Every verse in this repository was produced by a frontier AI model, with full provenance, and is released as a draft — not as a finalized translation. What you are reading is exactly what the AI produced, with the rationale for every decision visible alongside it. The repository is public so the process is inspectable from the first commit forward.

Contributing

Found a verse you'd translate differently? Open an issue using one of the templates under .github/ISSUE_TEMPLATE/. Engagement is welcomed from scholars, pastors, and lay readers. Our commitment is to respond publicly to every substantive concern.

Refreshing cached summaries after major revisions

The live reader caches chapter- and book-level summaries separately from the translation text itself. Missing summaries are filled automatically by the existing summary-prewarm workflow, but forced refreshes are manual on purpose. We do not automatically regenerate summaries on every push.

When a book has changed substantially, run:

scripts/refresh_book_summaries.sh "1 ENOCH"

Useful flags:

scripts/refresh_book_summaries.sh "1 ENOCH" --workers 20
scripts/refresh_book_summaries.sh "1 ENOCH" --chapters-only

Recommended policy:

• Definitely run it when a book's second pass is finished.
• Usually run it again after a third pass if major wording or framing changed.
• You can also run it after any other significant revision batch for that book.
• After the second and third pass, further summary changes should normally be small, so routine refreshes beyond that are optional.

Release cadence

The translation is built and released phase-by-phase, with each phase a full set of complete books (not partial books):

• Phase 1: Pauline epistles (Romans through Philemon)
• Phase 2: Gospels + Acts
• Phase 3: General epistles + Revelation
• Phase 4: Torah (Genesis through Deuteronomy)
• Phase 5: Former Prophets (Joshua through 2 Kings)
• Phase 6: Writings (Psalms, Proverbs, Job, Chronicles, etc.)
• Phase 7: Latter Prophets (Isaiah, Jeremiah, Ezekiel, Twelve)
• Phase 8: Deuterocanonical books (Tobit through 4 Maccabees, from LXX). Corpus complete as of 2026-04-20 (13 books, 6,337 verses, 98.9% high-confidence scan-adjudicated transcription). Translation drafting begins in Phase 9. See DEUTEROCANONICAL.md, REFERENCE_SOURCES.md (three-zone scholarly-source policy), and REVISION_LATER.md (deferred-source integration).

Tagged releases follow the vMAJOR.MINOR.PATCH convention. The first public release is v0.1-preview.

Directory structure

peoples-open-bible/
├── README.md            You are here
├── DEDICATION.md        Blessing and hope — to whom this work is offered
├── LICENSE              CC-BY 4.0
├── CHANGELOG.md         Phase-by-phase release notes
│
│   # Why and how
├── PHILOSOPHY.md        Why this translation exists, open-source posture, commitments
├── DOCTRINE.md          Theological commitments driving translation decisions
├── METHODOLOGY.md       Drafting and cross-check pipeline
├── REVISION_METHODOLOGY.md  How committed drafts are revised (reviser roles, revision criteria)
├── REVISION_ROADMAP.md  Revision-pass schedule and priorities
├── REVISION_LATER.md    Deferred-source integration (revisions queued for later passes)
├── REFERENCE_SOURCES.md Three-zone scholarly-source policy
│
│   # Corpus scope and per-book notes
├── DEUTEROCANONICAL.md  Strategy for the Deuterocanonical / Apocrypha books
├── EXTRA_CANONICAL.md   Strategy for the wider extra-canonical corpus
├── GREEK_EXTRA_CANONICAL.md  Greek-language extra-canonical scope
├── APOCRYPHA_PROVENANCE.md   Provenance notes for the Apocrypha
├── 2BARUCH.md           Per-book notes: 2 Baruch
├── 2ESDRAS.md           Per-book notes: 2 Esdras
├── DIDACHE.md           Per-book notes: Didache
├── ENOCH.md             Per-book notes: 1 Enoch
├── FIRST_CLEMENT.md     Per-book notes: 1 Clement
├── JUBILEES.md          Per-book notes: Jubilees
├── PRAYER_OF_MANASSEH.md  Per-book notes: Prayer of Manasseh
├── PSALMS_OF_SOLOMON.md   Per-book notes: Psalms of Solomon
│
│   # Public data snapshots (regenerable; see tools/)
├── status.json          Progress snapshot (tools/build_status.py)
├── revisions.json       Revisions index (tools/build_revisions_index.py)
├── book_metadata.json   Canonical book list, codes, ordering
│
├── schema/
│   └── verse.schema.json    JSON Schema for per-verse YAML
├── sources/             Vendored source texts (see sources/README.md)
├── translation/         Per-verse YAML (translation/nt/<book>/<chap>/<verse>.yaml)
├── tools/               draft.py, consistency_lint.py, validate_reader_corpus.py, …
├── docs/                Phase runbooks, cross-reference notes, design docs
├── outreach/            Correspondence with publishers (ESV, NLT, etc.)
└── .github/
    └── ISSUE_TEMPLATE/  Public disagreement and concern templates

Revisions dashboard maintenance

Progress statistics refresh after completed translation/review batches, not on a 30-minute timer. Unchanged inputs skip the full rebuild; laptop battery work is deferred. See batch refresh triggers and manual retry.