An atlas that asks a source to earn its credibility has to earn its own. This page states the rules this atlas follows when it decides what to call true, disputed, false or unknown, what a citation's reliability badge actually promises, how a disagreement becomes a public case with a published ruling, who does the work, and what it will not publish. It closes with a running account of where the atlas has gotten things wrong and fixed them, because a policy nobody can check is not a policy, it is a promise.

The Two-Part Confidence Badge

Every page here carries two labels near the top, and clicking either one opens the count behind it.

The first label says what kind of knowledge the page is recording: documented history, a tradition's own scripture or teaching, a living practitioner's system such as astrology or tarot, or a claim that scholars themselves still argue about. We use this label so that a practitioner's own system is judged on its own terms, not measured against a yardstick built for historical documentation. A tarot card's traditional meaning is not a worse kind of fact than a date in a history book. It is a different kind of fact, and we grade it as one.

The second label says how thoroughly the page's claims have been checked: how many independent sources back them, how reliable those sources are, how many of the page's claims are settled versus still debated among honest people, and whether anyone has formally challenged something on the page and, if so, what happened. We call the four levels Thoroughly Verified, Verified, Developing and Preliminary. None of this label is fixed. It moves as new sources are added, as old claims are re-examined, and as challenges are decided. If a claim we once marked as thoroughly verified later turns out to be wrong, we will say so plainly on that page, and we treat that correction as a serious event, not a routine edit, because the entire point of stating our confidence up front is that we do not get to quietly walk it back later.

Clicking either label opens the count behind it: how many claims, how many sources, how many still open questions. Nothing here is a black box.

Filing a formal challenge against a claim costs a reader nothing and moves nothing on its own. Only once we have actually reviewed it and accepted it into formal reconsideration does the page's confidence label reflect that something is under active review. This is deliberate: we do not want a page's reputation to be at the mercy of anyone who simply disagrees with it.

How a Claim Earns Its Status

Four words, and each one means something specific.

Every fact and every relationship in this atlas carries one of four statuses, chosen carefully because they are not interchangeable. Attested means the sources agree and the claim is established. Debated means real specialists hold competing positions, camps exist, and the entry names both sides rather than picking a winner it has no standing to pick. Debunked means the claim was checked and found false, and the entry says what the true account is instead. Unknown is the one people misread most often: it is not a softer version of debated. Debated is an argument you can go and read. Unknown means nobody has an answer at all, someone tried to settle it and stopped, and the gap is real.

An unknown is never left to sit by itself. Wherever this atlas records that it does not know something, it also records an open question naming what would actually resolve it and which field of expertise the answer needs, so the gap reads as an invitation rather than a shrug.

See every open question the atlas has recorded.

What the Source Tiers Mean

The tier says where a source comes from. It does not say the source is right.

Every citation on this atlas carries a reliability tier, shown as a badge beside it. Tier one is a scholarly reference or a peer reviewed work. Tier two is a reputable secondary source such as a museum, a university or an established reference body. Tier three is the tradition's own account of itself, cited as its own claim rather than as outside evidence. The current counts at each tier, and how many citations rest on each one, are published on How This Atlas Works.

The tier classifies where a work sits, never whether it holds up, and treating the two as the same thing is exactly the failure this atlas is trying not to repeat. A university press can be wrong. A tradition's own scripture is the right source for what that tradition teaches and the wrong source for whether the teaching is historically accurate, and the tier only ever answers the first question. What answers the second is a plain spoken explanation carried on every heavily cited source's own page: an audit of that source's track record, its editorial process, its known errors, and where later work has agreed with it or contradicted it. That explanation argues from the source's own history, never from its publisher or its category, because a publisher's name has never been a fact about whether one particular claim is true.

How Independence Is Measured

Ten citations that all copy one another are one witness, not ten.

A claim looks stronger the more sources back it, but only if those sources reached their conclusions independently. Much of what circulates as separate confirmation is one account, republished, summarised or popularised by everything that came after it. This atlas records that lineage directly: when a source is known to derive from, or to be a popular retelling of, an earlier one, the descent is written down as a link between the two source records, and every source is traced back through that chain to the earliest one it actually descends from. As of today, 98 such descent links have been recorded.

What this buys a reader is simple to state. A group of sources that all trace back to the same root counts as one line of evidence, not several, however many of them happen to be cited. This part of the record is young and growing, not finished, and it will only get more complete as more of that lineage is traced and written down.

How a Dispute Becomes a Public Case

A member disputes a claim. If it holds up, it becomes a public ruling with its reasoning attached.

Anyone with a verified account can dispute a specific fact or relationship on any page: naming what is wrong, proposing the correction, and giving a source that can be checked. That challenge is stored against the page and stays visible whether it is accepted or rejected, so the disagreement is part of the record either way.

A challenge that deserves a fuller look can be raised into a full appeal. An appeal is only accepted once it names the exact claim under dispute and is backed by real evidence, meaning a checkable source a reviewer can actually go and read, not an opinion or a restatement of the original objection. An appeal that clears that bar opens a case.

A case is worked by independently re-checking every source already behind the claim, not only the one sentence the appeal disputed, and each source examined is recorded as its own entry in the case file, whoever filed it. When a case is decided, the ruling always carries two parts: a short summary of what changed or did not, and the full reasoning behind it, in plain language, never withheld. The decision is one of four outcomes: the original claim is upheld, revised, overturned outright, or the appeal is declined. Only a legal problem or a claim touching a tradition's own restricted material is ever referred upward for a final call; every other case is decided and published on its own merits.

Disagreement does not end when a case closes. A reviewer, an independent re-checking team, or the person who raised the challenge can record a dissent from the outcome, and every dissent is published alongside the ruling it disagrees with, cited to a real source rather than left as an unsupported objection. A challenger who takes a claim all the way to a case, or who dissents from its outcome, is credited publicly by name. There is no anonymous option: raising a claim in this system also means being willing to be identified for it.

Browse every case the atlas has opened, decided or not.

Who Builds This

Disclosure, not apology.

This atlas is researched, drafted and entered by artificial intelligence research agents, working under the editorial rulings of a single human owner who is accountable for what is published. That is stated here plainly because it should be checkable rather than discovered, and because the standard this atlas holds its sources to, that credibility is earned by a visible track record rather than claimed on authority, has to apply to itself first.

What follows from that is not a lowered bar, it is where the real safeguards live. Every claim of substance is meant to carry a citation back to a source a reader can go and check for themselves. Every correction this atlas has made to its own published claims is kept on the record rather than quietly edited away. Disagreements with a ruling are published, not suppressed. None of that makes the atlas infallible. It makes the atlas's mistakes visible and its methods checkable, which is the only kind of trustworthy an automated reference work actually gets to claim.

What Is Deliberately Left Unpublished

A living tradition is described. Its closed material is not reconstructed to fill the description in.

Some of what this atlas records belongs to traditions that hold it closed: material passed on only to initiates, restricted by season, by ceremony, or by the authority of a teacher, or simply not meant for outsiders. Where that is true, this atlas says so plainly on the page rather than leaving a reader to assume the material was never found. It names that the restriction exists, describes in general terms what the material is for, and stops there. It does not reproduce the restricted material itself, and it does not reconstruct a plausible version of it from adjacent sources to fill the gap, because a plausible reconstruction of something a tradition keeps closed is not a neutral act of documentation, whatever the excuse for it.

The same restraint applies to material that already happens to be in print. That something was published once, sometimes by a researcher a community had no real power to refuse, is not treated here as a licence to republish it again. Specific examples of what is withheld, and why, are named on How This Atlas Works. A page being thin is not, by itself, evidence that its tradition is being reconstructed disrespectfully, and it is not evidence that the atlas simply has not gotten to it yet either. Both happen, for different reasons, and the running account below says which is which wherever the atlas actually knows.

The Atlas's Own Track Record

Read from the data on every visit, never written in by hand.

Zero facts on this atlas have been formally corrected to date: a page asserted one plain thing, a later check found it did not hold, and the page now says a different plain thing. Each correction stays visible on the page it corrects, in that entry's own voice, naming what was originally claimed, what checking it turned up, and the source that settled the question. Two examples on record: an early claim that a moon of Saturn was named for the titan Coeus, which no naming record actually supports and which the entry now states plainly; and an early claim crediting the Muse Thalia for the plant genus that in fact honors the botanist Johannes Thal.

Nine disputes have been carried all the way through the public case process described above. Of those, six upheld the atlas's original claim, three led to a revision, zero overturned it outright, and zero were declined at the evidence stage. zero member appeals have been filed under this process in total. This system is new, and a zero here is an honest zero, not a hidden one: it will grow as real disputes run their course, and every one will be published the same way.

Fifty-five traditions on this atlas carry a published note stating plainly that their page is short because the tradition's own knowledge is held in restricted custody or passed on orally rather than written down, not because the tradition matters less. Being an oral tradition does not, on its own, predict a thin page: some of the most thoroughly documented traditions in this atlas are oral ones that later attracted heavy outside scholarship, while some of the thinnest belong to traditions with millions of living adherents that outside writers simply never studied at length. The honest fix for that gap is documented material a community itself chooses to contribute, never invention to fill the space.

Every number in this section is read from the atlas's own records at the moment you load this page, not typed in by hand, so it cannot go stale without the atlas itself noticing.