← SAM home

The Scripture Analysis Method

SAM answers scripture-based questions from primary sources only, the Hebrew Masoretic text, two Greek New Testament traditions, the Septuagint, and a plain English baseline, never from commentary, tradition, or an AI model's memory. Every quotation you see has been mechanically verified, word for word, against the source corpus before it reaches you.

The sources

• Hebrew Old Testament, Masoretic text, Westminster Leningrad Codex, with every word tagged to its dictionary entry.
• Greek New Testament, two traditions, the Byzantine Majority Text (Robinson-Pierpont) and Tischendorf's 8th Edition, so textual variants can be checked against each other.
• The Septuagint (Brenton's English translation), the Greek Old Testament the New Testament writers often quote.
• World English Bible, the readable modern-English baseline.
• Strong's Greek and Hebrew lexicons, consulted whenever a question turns on what a word means; SAM never asserts a Greek or Hebrew word's meaning from memory.
• 1 Enoch (R. H. Charles, 1912), consulted only because the New Testament quotes it; it carries no authority of its own.

The canon boundary

SAM operates on the 66-book Protestant canon, selected by convergent criteria: apostolic proximity, doctrinal consistency with the verified historical core, cross-community verification, and citation by canonical authors. 1 Enoch is consulted (not canonized) because Jude quotes it as prophecy. The Apocrypha is excluded because no New Testament author cites it as Scripture.

Phase 1, Read the text honestly

1. Classify the question. Factual (what happened?), doctrinal (what is taught as binding?), practical (what is required?), predictive (what will happen?), or disputed (where do honest readers disagree?). Each type is answered differently, mixing them up is the most common source of bad interpretation.
2. Read the whole chapter, never a bare verse. Who is speaking, to whom, what occasioned it, what comes before and after. Every passage SAM cites was read inside its full chapter, enforced by the engine, not left to chance.
3. Identify the text type. Direct divine speech, law, narrative, poetry and wisdom (which use metaphor and hyperbole), prophecy (conditional or unconditional), or apocalyptic vision (heavy symbolism).
4. Check the command's own stated purpose. If a law says why it was given, that matters. But the purpose must be stated by the text itself, a purpose supplied by scholars or reconstruction is never enough to set a command aside.
5. Let scripture interpret scripture. Every other place the same word, concept, or theme appears is checked before anything external is brought in. For word questions, the original Hebrew or Greek is examined word-by-word with the lexicon.
6. Weigh authorial distance. A later text directly quoting an earlier one carries more weight than a later text applying it by analogy.

Phase 2, Establish authority

When texts conflict, or a command's standing is questioned: whose word is each text, and does it have standing to change what was established earlier?

• Tier 1, God's direct word. Explicit divine speech, in any book (Exodus, the whirlwind speech in Job, the “I am Yahweh” oracles of the Psalms and Prophets). Highest authority; cannot be overridden except by an explicit, stated revocation.
• Tier 2, Jesus' own words, read in full chapter context. Full divine authority; may clarify or fulfill Tier 1, but clarifying and revoking are different moves needing different evidence.
• Tier 3, The commissioned apostles' testimony: what the eleven (with Matthias) wrote (Matthew, John, 1–2 Peter) and what they proclaimed after Pentecost (their speeches in Acts). The tier tracks the testimony, not the person. Where one of the eleven is quoted as a character inside the Gospel narrative, his words sit at the narrative's tier: Peter's “You are the Christ” (Matthew 16:16) and Thomas's “My Lord and my God” (John 20:28) read as Tier 5, not because they are doubted, but because at that moment they are lines in someone else's narrative, not the apostle's own commissioned testimony.
• Tier 4, Paul & the later apostles: Paul, James, Jude, commissioned as apostles, but not eyewitnesses of Jesus' earthly ministry (Paul by a later vision). A notch below the eyewitness eleven.
• Tier 5, Everyone else: the narrators and non-apostle writers (Luke, Mark, the writer of Hebrews) and every Old-Testament figure speaking in his own voice. Valuable testimony that can clarify and apply, but cannot revoke what a higher tier established.

How authority transmits. One rule runs through the tiers: a person's authority attaches to what they wrote, and to testimony the text formally frames as theirs, never to being quoted inside someone else's story. Quoted speech carries the speaker's own tier only where the text itself marks it, and Jesus is the one editorially-marked case: the red-letter tradition marks his words, so they read at Tier 2 even inside Mark's or Luke's narrative. No such marking exists for anyone else, which is why the same apostle can be Tier 3 in his own letter and Tier 5 when a Gospel writer quotes him as a character in the story. The tier follows the speech act, not the name.

Two consequences of that rule, stated openly because careful readers ask about both. First, the Gospels-versus-Acts line: an apostle's speeches in Acts are formally framed as the commissioned witness the book exists to record, so they carry his testimony tier; the same man quoted mid-story in a Gospel is a character in another writer's narrative. Same literary mechanics, different formal framing, and the framing is what the rule reads. Second, demotion loses nothing: when Jesus himself responds to a confession (Matthew 16:17, John 20:29), the response is Jesus' own speech and enters the analysis at his own tier. The affirmed content is weighed at full authority through the affirmation itself, no “endorsement uplift” rule is needed, and adding one would replace a mechanical rule with a judgment call about what counts as an endorsement.

A note on attribution: placing a prophetic oracle (“thus says Yahweh”) in the top tier records the claim the text itself makes about who is speaking. SAM reads attribution from the text; it does not independently verify it, no method operating on the text alone could.

The core principle: silence is not revocation. A higher-tier command stands unless a text explicitly states it is revoked, never inferred from silence, omission, or absence of restatement. Two supporting rules: symbolic or poetic language never overrides a plain direct-speech command at the same or higher tier; and within the same tier, a later explicit statement on the same subject modifies an earlier one (Genesis 9:3 modifies Genesis 1:29, later silence modifies nothing). Whenever a lower-tier text is used against a higher-tier command, SAM names that move out loud instead of letting it happen quietly.

Phase 3, Deliver the answer, tagged by the kind of evidence

SAM does not tell you how confident it is, that would be an opinion you can't check. Instead it tags every answer with what kind of textual support exists, a claim you can verify yourself against the verses shown:

• Direct, a clear verse states it, and no higher-or-equal authority text overturns it. You can point to the verse.
• Inferred, no single verse says it; the answer is pieced together across texts. SAM names where the inference turns and what cuts against it.
• Contested, strong texts genuinely pull both ways and the text doesn't say which wins. Both sides and their sources are shown.
• Not in the text, scripture doesn't address it (including “is [a formula like the Trinity] explicitly taught?” when the wording is in no passage). SAM names exactly what the text is silent on.

For questions that list things (“which of the Ten Commandments does X break?”), SAM gives a per-item table, each item, the verses that bear on it, and its own tag, instead of one overall label. And it answers the question actually asked: if something conflicts with a command outside the set you asked about, SAM says so plainly and names the real command rather than forcing it in.

A further outcome, couldn't ground an answer, means SAM found no source passages that address the question, it says so rather than guessing. SAM never fabricates a tag.

When texts disagree

When the Hebrew and the Septuagint diverge on an Old Testament passage, both readings are shown, weighted toward whichever the New Testament quotes, if it does. When the two Greek New Testament traditions differ on a verse, the variant is flagged. Genuinely disputed questions get the strongest scriptural case for each side, then the tier system decides which, if either, the text favors.

What SAM never does

• Quotes a verse without having read its whole chapter.
• Lets a lower-tier text quietly override a higher-tier command without naming the move.
• Manufactures harmony between genuinely conflicting texts by softening one of them.
• Answers from memory when a source text can be checked directly, or invents when it can't.
• Treats silence, omission, or absence of restatement as revocation.
• Picks one source when sources conflict without telling you.
• Presents a denomination's tradition as the text's own position.
• Claims certainty the text does not supply.
• Treats what a story records as automatically being what is commanded.
• Collapses categories the text itself keeps distinct.

Follow-up questions

Ask follow-ups in the same session, “why that tier?”, “what about verse 12?”, and SAM continues from the same evidence it already examined.

Reproducibility

Every answer is stamped with the model version, corpus version, and framework version that produced it. Same pins, same verdicts: the exact wording of an answer may vary between runs, but the verdict, the cited verses, and their tiers are held fixed by the pinned configuration, and if any pin ever changes, a fixed test set of questions is re-run and reviewed before the change ships.

Sources & credits

Westminster Leningrad Codex + morphology (OpenScriptures, CC-BY 4.0) · Strong's Greek and Hebrew dictionaries (OpenScriptures; Hebrew CC-BY-SA) · Robinson-Pierpont Byzantine Majority Text · Tischendorf 8th Edition (MorphGNT) · Brenton's Septuagint (ebible.org) · World English Bible (ebible.org) · 1 Enoch, R. H. Charles 1912 (sacred-texts.com) · Versification mappings: Copenhagen Alliance (CC-BY-SA 4.0).

Ask a question →