This site is researched and drafted with the help of AI tools, and every primary or peer-reviewed source in it has to carry a permanent identifier — a museum registration number, a CDLI tablet number, a DOI, an ISBN, an archive identifier — that points at the item cited and can be followed by anyone. A page whose identifiers are missing or do not resolve is not published. This page explains exactly how that works, because you should not have to take it on trust.
Why this page exists
The value of this site is a single claim: that when we say an inscription says something, it does, and you can go and look. A site built on that claim cannot be vague about how its text is produced. So here is the whole process, including the parts that are uncomfortable.
Where AI is used
AI language models are used here for three things:
- Mapping the state of a question. Which positions exist on the dating of a text, who holds them, what the disagreement actually turns on, and where to go looking.
- Proposing candidate sources. Which artefact, which journal, which critical edition, which locus. Every one of these arrives as a candidate and nothing more.
- Drafting. Structure, first passes at the case for and the case against, and the mechanical work of turning notes into prose.
Where AI is not used
No factual claim and no citation on this site survives into published text on the strength of a model’s say-so. The verdict on a dossier is never delegated: it is a signed editorial judgement, made by a person, and it carries that person’s name.
The failure mode we are guarding against
Language models produce plausible citations that sometimes do not exist. A tablet number, a journal page, a year of publication, a position attributed to a scholar who never held it — this kind of error comes out of a model with exactly the same fluency and the same air of authority as a correct one.
On most websites that is a manageable defect. Here it would be fatal, for three reasons that compound. The asset of this project is verifiability, and a site whose pitch is “we checked the sources” does not survive a reader who checks and finds nothing. The audience for a documentary dossier is, by definition, trained to check footnotes. And in a field occupied by institutions with reputations to protect and apologists with positions to defend, one invented citation in an unfavourable verdict is ready-made ammunition.
So the whole workflow is built around one rule: the model proposes, and nothing reaches the site until the citation has been resolved against the institution that holds the item. Not summarised, not recognised — resolved, to a permanent identifier that you can follow yourself.
What “resolved” means here, and what we do not claim
Every primary and peer-reviewed source on this site has to carry a permanent identifier issued by whoever holds the item: a British Museum registration number, a CDLI P-number for a cuneiform tablet, a DOI for an article, an ISBN for a book, an accession or archive identifier for anything else. The identifier is the thing that can be checked by a stranger, which is the only kind of check worth publishing. A source that cannot produce one is recorded as NO_IDENTIFIER and does not count towards anything.
Let us be plain about what this is not. Nobody here travelled to a museum. This site does not claim that a member of staff stood in front of a tablet, and you should distrust any site that claims it without saying who and when. What we claim is narrower and testable: the identifier printed beside each source resolves to the item described, and you can follow it from the page.
Three things are recorded for every source, and all three are required: the permanent identifier, the address at which it resolves, and the date the resolution was last confirmed. The date is stamped by the software, not typed in, so that it cannot be filled in from memory. Changing the identifier resets the date, because a different identifier is a different item.
The resolver queries each identifier against the institution that holds the item — Crossref for a DOI, Open Library for an ISBN, the archive’s own metadata endpoint, the museum’s object page — and compares what comes back with what the page claims. A page that answers with the wrong title, or the wrong author, or answers 200 without containing the item at all, does not count as resolved. Everything published is re-tested weekly, so that a record which is moved, renumbered or withdrawn shows up instead of rotting quietly in a footnote. When a live link dies but a dated snapshot exists, the source stays publishable and the page says it is reading from the snapshot.
The result for every source on this site — identifier, state, date of last resolution, snapshot — is public at the verification log, in HTML and as JSON. That page is the whole point: it lets you audit the site without asking us for anything.
A primary or peer-reviewed source whose identifier is missing or does not resolve cannot be published, and no dossier that cites it can be published either. This is enforced by the publication gate, in software, not by discipline. There is no button that publishes a page anyway; we did not build one, deliberately, because a gate with an override is a suggestion.
The provenance block
At the foot of every dossier you will find a provenance block: the author and their credentials, the publication date, the date of last review, how many sources sit in each evidence layer, and the identifier and resolution date of each one. None of it is written by hand. It is generated from the database, which is why it cannot flatter us. If a source is linked but its identifier is unresolved, the block says so in plain words.
What limits publication here
There is no monthly quota here. A quota is a promise about effort, and effort is not something a reader can check — so this site promises something testable instead.
A dossier does not publish unless at least two primary or peer-reviewed sources are attached to it and every one of their identifiers resolves. Resolving means software fetched the record, compared the metadata that came back against what the citation claims, and stored the result with a date. Everything published is re-tested weekly, because links rot without announcing it. A page whose citation stops resolving returns to draft until it resolves again, and there is no button that publishes anyway.
That rule is stricter than a number, because a number can be met by writing four weak pages and a rule like this one cannot be met at all until the citations hold. It is also the rule the software actually enforces, which is the only kind worth publishing. Every source, its identifier, its state and the date it was last checked are listed in the verification log, along with a monthly count of every citation that has ever resolved to the wrong thing.
Assistance does not turn a six-hour dossier into a forty-minute one. It compresses the drafting and the mapping; it does nothing at all to the part where every claim has to be tied to an identifier that resolves, and the ones that cannot be tied have to be cut. Honestly stated: a dossier that would take four to ten hours takes three to six, and the share of that time spent on citations goes up, not down.
If you find an error
Write to contact@scriptureandrecord.com with the page and the problem. Corrections are published on the corrections page with a date, and the affected page is annotated. We would rather carry a visible list of things we got wrong than a quiet reputation for never having been caught.