Inclusio.

What this site will not do

This is a tool for reading the Bible. A confident sentence about God’s word that turns out to be false does more damage than an empty margin ever could. So the rules below are constraints on the software, not aspirations for it — and several of them cost real features.

The four rules

1. Scripture is copied, never composed.

Every character of every verse is read out of a source file on disk and passed through untouched. No part of this site generates, paraphrases, completes, modernises or “corrects” a word of the biblical text. No language model is involved at any point in producing the text you read. Every book declares its verse count per chapter in advance, and the build refuses to run if the parser produces anything else — 48 verses of Jonah, 85 verses of Ruth, 167 verses of Esther, 21 verses of Obadiah, 47 verses of Nahum, 25 verses of Philemon, each in unbroken order. A program that silently drops scripture is worse than no program.

2. Every claim in the margin names how to check it.

Each note carries a basis, and there are only four permitted: an observation about the text you can verify by reading it; a value computed at build time from a named dataset; a footnote written by the translators themselves; or an open question where scholars disagree. There is no fifth basis. In particular there is no “well known” and no “commonly understood.” A claim that cannot carry one of the four does not appear.

BasisMeaningNotes
textVerifiable by reading the passage32
computedDerived at build time from a vendored dataset12
translatorsQuoted from the WEB committee’s own footnotes7
disputedNamed positions, no winner picked3

3. Where scholarship is unsettled, the page stays unsettled.

The site does not flatten a live disagreement into a clean answer because a clean answer looks better. It names the positions and leaves them standing. You are trusted to hold two of them at once.

4. Nothing here is a substitute for the thing itself.

Information about a text and relationship with its author are different things, and the first can quietly counterfeit the second. That is the real failure mode of a project like this one — not a wrong distance, but a reader who ends up feeling well-informed about God instead of close to him. Every chapter therefore ends in a stop rather than a next, and the way onward does not appear for a few seconds. That delay is not a bug.


A worked example: the data we threw away

The source file for the World English Bible ships a Strong’s concordance number attached to nearly every English word — 1,077 of them in Jonah alone, across 173 distinct numbers. That would make a word-study feature almost free. We do not use it, because it is wrong.

The tags are machine-aligned. In Jonah, 164 words carry H3068, the number for the divine name — but only 26 of them are actually the word “Yahweh.” The other 138 (84%) are ordinary English words the aligner spilled the tag across:

“to”, “the”, “But”, “of”, “He”, “a”, “he”, “sent”, “on”, “and”, “was”, “that”

Had we shipped the obvious feature, this site would have told you that the English word “a” is the personal name of God. So the tags are stripped out of the text at build time and never rendered. Word notes are hand-checked one at a time instead, which is slower and produces far fewer of them. That trade is the whole point.


A second example: the map that will not pin Tarshish

Jonah’s flight is the most quotable fact in the book — he was sent to Nineveh and paid for a ship to Tarshish. The map on the Jonah plate draws Joppa and Nineveh as points, because the vendored OpenBible rows for both cite the verse that names them and give an exact coordinate. Tarshish does not get that treatment:

FromTowardDistanceBearing
JoppaTarshish 3,779 km (great-circle) —
JoppaTarshish — 290° WNW

We cannot put a confident pin in Tarshish, because nobody knows where it was. Our own two source files disagree with each other: OpenBible’s places.txt has no numeric coordinate for that row at all; merged.txt supplies one, in southwest Spain. The map draws that point as an open ring — “disputed,” never a pin as certain as Nineveh’s — and adds the one thing the dataset actually supports without naming a place: a bearing of 290° (WNW) from Joppa, toward Tarshish. We do not draw other proposed identifications (Tarsus, Sardinia), because the vendored dataset identifies Tarshish with neither. It does carry Tarsus, as a separate place of its own, and that point is not west of Joppa at all — so this site no longer says “every candidate lies west”, which it once did. See the known gap below.

So the distance above is “the distance to one disputed identification of Tarshish,” and the site says so wherever it appears. What survives the uncertainty is what the text itself says: he was sent to Nineveh, and paid for a ship to Tarshish instead (Jonah 1:2–3).


Known gaps, stated plainly

If you find an error

Errors in the apparatus are bugs and are treated as bugs. Errors in the scripture text itself would be a serious defect — the text ships with a recorded SHA-256 checksum precisely so that any drift from the published source file is detectable rather than a matter of trust. Both checksums and retrieval dates are on the Sources page.