How a claim is verified

Published deliberately, all of it. A registry that asks people to trust its verdicts should say what those verdicts are based on and who reaches them — and knowing the checklist in advance helps submitters take the right photos the first time.

The process

Proof that the card is in front of you

Before a single photo is taken, the site issues a random phrase and a simple figure to draw. Both have to be handwritten and appear in the same frame as the card. Because they are generated at the moment you start — and never shown as a picture to copy — a photograph lifted from someone else's pull video cannot contain them.All of this is optional, and that is a deliberate trade the registry makes rather than an oversight. Only a photograph of the front is required to file. Asking for the full set was costing us serials — people started a claim, saw the work, and gave up — and a serial nobody reports is worse for a public register than a thin claim a reviewer can chase.What that costs is stated on the record rather than absorbed. Every copy page carries a badge saying what proof was actually taken: phrase and figure, phrase only, or no proof of possession at all. Nothing thin is ever presented as though it were thorough, and a claim with no proof shot needs two reviewers to agree rather than one.

Automated checks — advisory only

A handful of deterministic checks run on filing: whether the serial is already claimed, whether the photographs have appeared before (both exact copies and re-cropped or recompressed ones), and whether the images are what they claim to be. These produce hints beside the evidence.Nothing automated can approve anything, at any confidence level. There is no code path from a passing check to a confirmation, and there is not going to be one — if a check were ever good enough to approve unsupervised, the right use of it would still be to make a person faster.

An expert reviewer works the checklist

Every claim is read by an experienced reviewer — someone who handles these cards, knows how the foil moves, and knows what the print looks like under magnification. They work through the published checklist below, item by item, recording a verdict and a note against each one.The decision is attributed to a named person, dated, and stored against the exact version of the checklist they judged by. When the standards tighten later, a claim approved in March still reads as approved against March's rules rather than being quietly restated under June's.

Two reviewers on every public claim

Every claim filed through this site needs two independent sign-offs from two different people before it is confirmed. The reviewer who gives the first is barred from giving the second. Between the two the claim sits visible as pending and is on nobody's public record.That is the other half of the bargain in step 1. Lowering what a submitter has to send raises what the registry has to do with it, so the second reader went where the evidence is now thinnest — a stranger, an email address and possibly a single photograph.Records entered directly by an administrator are the exception, and they are labelled as such on the copy page, permanently, together with whatever account of their origin the admin gave. Those take one approval rather than two — but never from the administrator who entered the record. Somebody else always has to be the one who publishes it, and the database refuses the write otherwise. The page never claims proof of possession was taken when it was not. Marquee cards override all of it and cost two approvals from either route.

Competing claims are not resolved by picking a winner

When two people both make a credible case for the same serial, that is itself evidence something is wrong, and choosing between them would launder the problem into a false confirmation. Both claims are published side by side with their evidence, the serial is marked disputed, and neither is confirmed.A confirmed copy can be pushed back into dispute by a later claim. Confirmation is not permanent.

The checklist itself

This is the working list reviewers use, exactly as they see it. It changes as the set becomes better understood — every edit is a new version, so no past decision is retroactively rewritten. Items marked blocking must pass before anything is confirmed.

Evidence integrity

The handwritten phrase matches the issued one exactly.blocking
The handwritten phrase matches the issued one exactly. Handwritten, not typed or digitally overlaid.
The date on the slip falls between the draft's creation and submission.
Same surface, lighting, and hands across all shots.
Same surface, lighting, and hands across all shots. A proof shot taken somewhere else is the most common sign of borrowed images.
Matching whitening, scratches, and centering identify it as one card, not several.
EXIF present and internally consistent; capture times in order; no editing-suite signature.
Dimensions and compression artifacts consistent with a camera, not a screen grab or video frame.

Serial and catalog match

The serial is readable and within `1..total_serials`.blocking
The same serial appears in `card_front` and `serial_closeup`.blocking
Card name and set code match the claimed catalog entry.
Which run this card is from — Americas or Europe — is established.blocking
Which run this card is from — Americas or Europe — is established. The two differ **only** by an `E` after the serial, and submitters routinely omit it. Read it off `serial_closeup`; do not assume Americas because no suffix was typed.
Not already `confirmed` against a different owner **in the same region**.blocking
Not already `confirmed` against a different owner **in the same region**. `#042` and `#042E` are different cards and never conflict with each other.
The claimed pull date is on or after the set's release.

Physical authenticity

The dark inner layer is visible at a cut edge.
The dark inner layer is visible at a cut edge. A solid-white edge is a strong counterfeit signal.
The holo shifts coherently as the card tilts.
The holo shifts coherently as the card tilts. Flat printed-on "shine" that does not move is the single most reliable giveaway — this is why `angled_foil` is a required shot.
Genuine cards show a regular offset-print dot pattern.
Genuine cards show a regular offset-print dot pattern. Laser banding or inkjet spray indicates a fake.
Font weight, kerning, and edge sharpness against a known-good reference.
Font weight, kerning, and edge sharpness against a known-good reference. Counterfeits commonly get the name font subtly wrong.
Copyright text and set code formatting and position.
Colour accuracy and sharpness of fine linework.
Even borders, consistent corner radius, no delamination.

Provenance and possession

If a source link is given: it exists, shows *this* serial, and predates the submission.
If filed for someone else, the actual puller is named and ideally corroborated.
Whether the submitter claims ownership.
Whether the submitter claims ownership. Only a claim creates an owner record and a custody event.