VERIFICATION PROTOCOL

A score enters ENDSCREEN as a photograph, not as a number typed into a form. An operator finishes a run, photographs the end screen off the cabinet or the display with the ENDSCREEN field app, and the app reads the score from that image. The image travels with the score for the rest of its life on the archive: it is what the board shows when you open a run's evidence, and it is what every later submission is compared against.

A run may also carry a replay. Where one is attached, the board marks the score with a ▶ VOD tag, and the replay is linked from the row. A replay is additional evidence for a number that was already photographed — it is never a substitute for the capture.

Every capture is perceptually fingerprinted and compared against every other capture already on file. A perceptual fingerprint describes what an image looks like rather than what bytes it contains, so re-cropping, rescaling, recompressing or lightly recolouring a screen does not produce a new image as far as the comparison is concerned.

When a submission matches evidence another operator has already filed, the archive flags it. The row carries a ⚠ DISPUTED marker wherever it appears, and that marker travels with the capture — it is visible on the board, in the operator's own log, and in the trending feed on the front page.

The flag is a statement about the evidence, not a verdict on a person. Two operators can photograph the same screen in the same room; a disputed capture is one the archive cannot tell apart from another, and it is published as exactly that.

A board is a title, not a machine. Each operator's best sealed capture for a title is ranked against every other operator's best for that title, across all of the hardware ports the title exists on, rather than splitting one game into a leaderboard per platform. Where a title runs on more than one port, the board can be filtered to a single port — but the page, its canonical url, and its ranking are the whole title.

A capture that is disputed still appears and still belongs to the operator who filed it. What it cannot do is quietly become the record: the archive tracks the best undisputed capture separately, which is the one the system treats as an operator's standing on that board.

A circuit is a time-boxed competition on one title. Entering debits a fixed buy-in from the entrant's wallet, and every buy-in is held in escrow by the system until the circuit closes. The pot is the buy-in multiplied by the number of entrants — it grows with every operator who joins, and it is paid to whoever leads the circuit at close.

If a circuit closes with no qualifying score on it, nothing is kept. Every entrant is refunded their own buy-in in full.

What qualifies depends on the circuit's mode, which is stated on every circuit row and detail page. LIVE RUN counts only captures logged inside the circuit's own window; ARCHIVE SYNC counts each entrant's all-time best for the title. Hardware scope is stated the same way: a circuit either counts every port of a title or names the ports it counts.

It will not present unverified data as verified. When a board, a feed or a circuit list cannot be read live — because the source is unavailable, or because the data is a placeholder — the surface says so in place of the figure, labelled FIXTURE or LOCKED. No such figure is ever published to a search engine as a real score: the structured data behind a board is emitted only for boards that read live, and the sitemap lists only boards holding at least one sealed capture.

It will not rate a game. ENDSCREEN ranks runs, not titles. There is no score out of ten here and no aggregate rating in the markup, because inventing one to win a row of stars in a search result would be a claim nobody made.

It will not quote an apex it cannot stand behind. Where a board's prose, its page description or its social card names a current record holder, that figure is read from live captures. A board with no sealed captures says it has none.

Every capture on the archive is inspectable. Open a board, expand a row, and you can see the evidence the ranking was built on — the photograph, the replay where one exists, the capture time, and the hardware port it was logged against. A score that cannot survive that inspection is not one the archive is interested in defending.

The fingerprint is automatic and continuous: it runs against every new submission, so a capture filed today is compared against the whole archive as it stands today, not only against what existed when the board was built.