Editorial standards
A value is a commitment held by a curator. A criterion is a test applied to an artifact. This page states the criteria, because a value that cannot be failed does not constrain anything — and because a standard nobody outside the project can hold it to is not a standard.
Criteria are numbered so they can be cited in review notes and in the revision history of an entry. Version 0.1, and it will change; when it does, the change is recorded.
Criterion neutrality
Traditions disagree not only about conclusions but about what makes a claim credible in the first place — apostolic proximity, conciliar authority, scriptural sufficiency, continuous tradition, magisterial teaching, prophetic revelation, rabbinic consensus. Adopting any one of those as the library’s measuring apparatus would decide the outcome before the evidence was read, while every individual statement in the entry still looked defensible.
| C-1 | No tradition’s framework supplies the measuring apparatus for another’s. Where traditions disagree about what makes a claim credible, the library documents the disagreement. |
|---|
| C-2 | Sequence, not verdict. Chronology and attestation are documentary findings. What a chronology means is an interpretive claim and belongs to the traditions, in their own voices. |
| C-3 | Providential readings are excluded from entry content. A claim that God permitted or arranged a historical development cannot be checked against a source, and appears only attributed to a party who holds it. |
| C-4 | Inherited vocabulary is inherited framing. Terms that encode one tradition’s evaluation of another are used only when attributed to the party using them. |
Record and inference
| C-5 | Every entry separates what the record establishes from what interpretation infers, structurally — distinct sections, not careful wording. A reader must be able to tell which is which without trusting the curator. |
| C-6 | Contested readings are named as contested, with the competing readings stated. “Most commentators hold X” is a fact about a scholarly community, not an argument. |
| C-7 | Silence is reported as silence, and arguments from silence are labelled wherever an entry relies on one. |
Steel-manning
Every claim an entry answers is stated first in its strongest form, in the vocabulary its holders use. Where that makes an entry harder to read, the remedy is a definition, not a simplification that would weaken the position while appearing to state it.
| C-8 | The steel-man block is physically first in the reading column. A rendering commitment, not only an editorial one. |
| C-9 | Sourced to a named proponent or institution that actually holds the position. Not a composite, not a reconstruction, not “critics say”. |
| C-10 | Stated in the proponent’s own technical vocabulary. |
| C-11 | A steel-man fails if it states the position in terms its holders would reject, substitutes a more answerable version, buries the response inside it, characterises the proponent’s motive rather than argument, or is sourced only to the position’s critics. |
Objections
| C-12 | An objection to an inference lives inside the entry that draws the inference, before the response. Not in a separate entry, not in a related-links list, not in a later section. It attaches if, read in the ordinary way, it weakens the inference — regardless of which author, century, genre, or tradition it comes from. |
| C-13 | Difficulties are disclosed before a critic raises them. The test is not “would a reviewer find this” but “would a competent opponent find this”. |
| C-14 | Objections carry a strength rating and the rating is honest. Rating a strong objection weak in order to dispose of it is the most damaging single failure available to this library, because it corrupts the instrument readers are being asked to trust. |
What a text supports, and what it derives
Entries state which of three levels is being claimed, because collapsing them is the commonest way a religious argument overstates itself.
| Derivable | The position follows from the text on its own terms. The text must compel the reading against serious alternatives. |
|---|
| Available | Consistent with the text and supported at identifiable points, but requiring premises the text does not supply. Both the points and the added premises are named. |
| Read into | The position is imported and the text accommodates it. The entry says so. |
| C-16 | Revelatory claims are labelled as revelatory, always. A claim resting on revelation is never presented as though philology, archaeology, or textual criticism produced it — with particular force where a revelatory claim redefines a term against its attested usage. |
| C-17 | Where a position is available but not derivable, the defensible claim has this shape: the record has certain features; the position accounts for them at least as well as the alternatives; the position is offered as revelation and the record is offered as consistent with it. Not: the record establishes the position. |
| C-18 | A tradition’s own translation is not evidence for the claim under examination. It may be cited as that tradition’s reading, labelled as such. |
Sources
| Primary | Evidence of what a source says. Critical editions, manuscripts and facsimiles, official magisterial and conciliar texts, canonical scripture in the original languages, liturgical texts, first-hand documents. |
|---|
| Reference | Evidence of attested usage. Standard lexica, concordances, critical apparatus. |
| Scholarship | Evidence of what a scholar argues. Peer-reviewed monographs and articles, recognised academic reference works. |
| Tradition-internal | Evidence of what a tradition holds. Catechisms, confessions, official statements, authoritative teaching. |
| Disqualified | Aggregator sites, general-reference encyclopedias, AI summaries, apologetics and counter-apologetics blogs, and aggregator reproductions of primary sources. |
| C-20 | Claims about a tradition are cited to that tradition’s own sources. Never to its critics, never to another tradition’s characterisation of it. |
| C-21 | Aggregator copies of primary sources are not primary sources. The critical edition or the facsimile, not a site reproducing it. This governs links as well as citations. |
| C-22 | Popular claims with no traceable primary source are retired, not repeated. Circulation is not attestation, and a claim repeated across traditions is not thereby corroborated. |
| C-23 | Contested dating is stated at the point of use, not in an endnote. A text that cannot be securely dated before the material it is supposed to illuminate cannot be cited as background for it. |
One consequence worth naming: the library does not link to hostile polemic, including where it reproduces a primary source. Linking is a stronger form of naming than citing, and the underlying claims are handled on their merits in Common Claim entries instead. The absence is a policy, not an evasion.
Confidence ratings
| Strong | Supported by primary evidence and not seriously contested among scholars who disagree about the surrounding questions. |
|---|
| Substantial | Well supported and defensible; serious objections exist and have serious answers, which are stated. |
| Contested | Serious scholars hold incompatible positions and the evidence does not decide between them. The entry does not resolve it. |
| Weak | Available but poorly supported; the objections are stronger than the responses. Stated as weak, with the reason. |
| Insufficient | Cannot presently be rated. The specific outstanding item is named. |
| C-25 | Weak arguments favouring the curator’s tradition are rated weak by name. This is the criterion by which the library’s symmetry claim is tested. If it is never exercised, the claim is decorative. |
| C-26 | A rating is auditable: the evidence supporting it is present in the entry. A rating a reader cannot check is an assertion of authority, which this library does not claim. |
| C-28 | Where an entry’s sensitivity setting suppresses the display of a confidence rating, the suppression is shown to the reader as a named state rather than left as a silent omission. |
Failure modes, named so they can be caught by name
| Criterion-level asymmetry | One tradition’s standard of credibility applied to all. |
| Sequence as verdict | Chronology converted into a judgement of legitimacy. |
| Unlabelled providential reading | God’s purposes asserted about a historical development. |
| Inherited framing | Evaluative vocabulary from one tradition used in the library’s own voice. |
| Consensus as argument | “Most commentators” substituted for evidence. |
| Weakened steel-man | A position stated in terms its holders would reject. |
| Detached objection | An objection relocated out of the entry it threatens. |
| Deflated rating | A strong objection rated weak to dispose of it. |
| Derivation overclaim | Available presented as derivable. |
| Revelation as philology | A revelatory redefinition presented as a lexical finding. |
| Circular translation | A tradition’s own translation cited as evidence for that tradition’s reading. |
| Circulation as attestation | A widely repeated claim treated as sourced. |
| Anachronistic background | An undated or later text cited as background for earlier material. |
| Tu quoque | A difficulty answered by pointing at another tradition’s difficulty. |
How AI is used, and what it does not confer
Drafting is AI-assisted. Verification is not. Every entry carries three independent markers — drafted-with-AI, reviewed-by-human, sources-verified — and all three must be satisfied before an entry is presented as anything other than a draft. The first is disclosed permanently and never removed.
AI accelerates work that would otherwise require an institutional team. It does not confer authority. This library’s authority comes from primary sources and from human reviewers, which is the reason the reviewer terms are written the way they are.
Sacred boundaries
Some material a tradition holds as belonging to a specific context is not catalogued here at all. Specific elements of Latter-day Saint temple ordinances are excluded on this basis; the same principle applies to the seal of confession, to particular sacramental and liturgical details, and to the parallel boundaries of every other tradition. Each tradition defines its own boundary. If an entry has catalogued something that should not have been, saying so is enough — the correction does not require argument or justification.