Missing beats wrong: how a match page earns its facts

Twelve attendance figures in our archive are deliberately frozen mid-disagreement. That's not a bug. It's the whole philosophy.

Every match page on this site is built on one editorial rule: a missing fact is a shrug, a wrong fact is a scandal. The archive would rather show you an em-dash than a number it cannot defend. This piece is about what that rule looks like in practice, because the practice is stranger and more interesting than the slogan.

Two elevens or nothing

A starting line-up publishes only when the source page carries exactly eleven names a side. Ten names is a broken page; twelve is a table we misread; either way, publishing it would put a wrong line-up under a real match, so the gate refuses the whole sheet. Before any sheet publishes at all, the page's own account of the goals must agree with an independent record of the match — scorer by scorer. In our measurements, pages that pass this cross-check disagree with the independent record 0.00% of the time; unchecked pages err at 4.17%. That difference is the entire reason the gate exists.

The twelve frozen attendances

Sometimes two good sources simply disagree. For twelve matches — all European qualifiers from the 1980s — our two best sources state two precise, different attendance figures. Not a rounded estimate against an exact count, where the precise number would win, but two exact counts: for Italy–Sweden 1983, 73,438 against 60,086. For Luxembourg–Scotland 1987, 2,022 against 1,999. For Italy–Sweden 1987, 59,046 against 59,045 — a disagreement of exactly one human being. We cannot know which is right, so the archive holds the incumbent figure, prints the dispute in its own records by name, and waits for a third source to settle it. Twelve numbers, frozen honestly, in an archive of 28,434 sheets.

Credit is a load-bearing wall

Much of the deep historical record exists because individual researchers spent decades compiling it. Where our sheets carry facts from the RSSSF's compilers, the compiler is credited by name on the page — that credit is a licence condition, and our loading rules treat it as unremovable: no later update, from any source, is permitted to delete an attribution. The archive's 15,723 recorded bookings and dismissals carry their sources the same way, lane by lane, and when two lanes disagree about a card, neither publishes until the disagreement is resolved.