The Material in Hand
Start with the Word, Not the Translation
The About page describes how I study. This page describes something different: how the project works — the machinery I use, with Claude, to scrutinize Scripture and to scrutinize this site's own content before you read it.
It begins with the material itself. I keep the texts in hand — the full King James Bible, Strong's Concordance with its original-language definitions, the Hebrew text where the work requires it — indexed locally, searchable with code. When I want to study a word, I do not search the English word. I find the Hebrew or Greek word first, and pull every passage that carries it — regardless of how the translators rendered it in each place.
Why that order matters is visible in the very first verses of the Bible:
"When God began to create the heavens and the earth, the earth was complete chaos, and darkness covered the face of the deep, while a wind from God swept over the face of the waters."
Genesis 1:1–2 · NRSVue
Even the first three words are a translation decision: the NRSVue's own footnote offers the traditional "In the beginning God created" as the alternative, and its predecessor the NRSV opened differently again. The opening clause of the entire Bible is contested rendering territory — which is precisely why the project anchors on the Hebrew rather than on any one English voice.
| Hebrew word | In Genesis 1 | Elsewhere in English Bibles |
|---|---|---|
| אֶרֶץ ʾereṣ | "earth" | Also rendered "land," "ground," "country" — the same word names the planet, the soil, and the promised land. An English search for "earth" silently misses most of them. |
| שָׁמַיִם šāmayim | "heavens" | Grammatically dual in form, and rendered both "heaven" and "heavens" — sometimes within the same book. Whether it means sky, the visible expanse, or God's abode is carried by context, not by the English. |
| תֹּהוּ וָבֹהוּ tōhû wābōhû | "complete chaos" | The KJV's "without form, and void." Two different English Bibles give you two different mental pictures of verse 2 — the Hebrew pair is what they are both reaching for. |
| רָקִיעַ rāqîaʿ | "dome" (1:6) | The KJV's "firmament." An expanse, something hammered out. The English words don't even agree on what kind of object it is — the Hebrew is the stable thing to study. |
Four words inside the same opening passage, each translated differently across — and sometimes within — English Bibles. This is not a complaint about translators; it is the reason for the rule. The original word is the index. The English is a rendering of it. So when this site studies "earth," it is studying earth (אֶרֶץ | ʾereṣ | ) — every occurrence, every rendering — not the English word "earth."
The Resources
Claude helps me build and retrieve the working library: the public-domain King James text, Strong's Concordance with original-language definitions, classic Bible commentaries, and current online material — the history, culture, language, and archaeology around the text. The point of holding these locally and indexed is speed with verification: a claim can be checked against the actual source in the same sitting it is made, and the pivotal references get re-verified against independent tools before they reach a page.
While building this very page I learned that there is archaeological evidence of Balaam son of Beor — the seer of Numbers 22–24. The Deir ʿAlla inscription, found in Jordan in 1967 and dating to roughly the eighth century BC, names "Balaam son of Beor, seer of the gods" as the figure of its own visionary text.
Note what this is and is not. It is not proof of the events in Numbers. It is an ancient, non-biblical text that knows the same name, the same patronym, and the same vocation. I record it because it is genuinely interesting and because it is context — not because it closes an argument. That discipline, applied every time, is the method.
Definitions: Saying the Same Thing Every Time
The project keeps a working definitions file — English terms, original-language terms, and the contested theological words — so that when a study says "evil," "sin," "moral," or "good," it means the same thing the last study meant, and the same thing the Hebrew or Greek underneath it can actually bear. Every definition carries a source label, because a definition is only as strong as where it comes from, and the labels keep five very different kinds of claim from blurring together:
| Label | What it tells you |
|---|---|
| Lexicon | Established by Hebrew/Greek lexicography or standard reference works — the firmest tier, and still verified against the text. |
| Lee | My working philosophical or theological position — held and argued, but flagged so it is never mistaken for a neutral starting point. Engaging it honestly means testing it, not adopting it. |
| Project method | A convention adopted for this project — authoritative here, not universally. |
| Contested | Scholars, traditions, or the texts themselves disagree. The file does not settle it, and content built on it must show both sides. |
| Site article | A fuller treatment exists on a published page; the definition summarizes it, it does not replace it. |
Just as important: the definitions allow some things to remain undefined. The file ends with a standing list of questions it is forbidden to settle — genuinely open matters where any tidy definition would be a theology smuggled in as vocabulary. Where Scripture leaves a term's edges dark, the entry records the openness instead of resolving it by fiat, and a draft that quietly settles one of those questions gets flagged, not published. The file also records its own corrections: when a definition I held collapsed under challenge, the entry says so, says when, and says why — so a later session cannot quietly re-derive the broken version.
Claude as Skeptic and Scholar
The theological positions on this site are mine, formed before AI is consulted. What Claude does with them is the opposite of agreement: I provide the Scripture, the evidence, and the narrative, and I ask Claude to critique all of it the way a skeptic would and the way a scholar would — name the gaps, the overreaches, the counter-texts, the readings I haven't considered, and the places where my conclusion outruns my evidence.
And when a critique lands — in either direction — it is written down. Every correction is recorded so it is not repeated. The record includes Claude correcting me, me correcting Claude, and the text correcting both of us. Some of those corrections moved real conclusions on this site; the studies they produced say so on the page. A standing file of past mistakes is read at the start of every working session, which means the project's scar tissue does what scar tissue should: it stops the same cut twice.
“A Claude that agrees easily is useless here.”
The working agreement, verbatimFrom Draft to Published
Once a draft exists, I sit with it for a few days. Distance is cheap and it works; a draft that still convinces me on Thursday has survived its first reviewer. Then, before anything publishes, it goes through four checks:
Are the article, its claims, and its evidence coherent and properly scoped? Does the conclusion follow, and does it claim only what the evidence covers? This is the graded check — the machinery is documented on How We Grade Ourselves, Blind.
Are the historical, linguistic, and archaeological claims consistent with secular and historical scholarship? Where a piece takes a minority reading or departs from consensus, the departure must be named on the page — checked against scholarship, not conformed to it.
Are the theological claims logical and internally consistent — with the piece's own stated bias, with the definitions file, and with what the rest of the site has committed to? Open questions must stay open; no draft gets to quietly settle one.
Does the site need this piece? It is scored for whether it adds a topic, a method, or a story the site doesn't already have — and whether its claims are conservative where they should be. Redundant or overreaching pieces get revised, merged, or cut.
Behind the Scenes
All of it leaves a paper trail. Every working session that touches content ends with a written record: what was claimed, what was cited, what was challenged, what changed. Every grade is logged with its scores and the exact version of the file it graded. Every correction joins the standing file. Recommendations — taken and not taken — are documented, so a future reader of the record can see not just what the site says, but what it declined to say and why.
None of that makes the conclusions right. It makes them checkable — which is the most any honest method can promise, and the least this site should offer the reader it hopes to earn.