Research · Applied
The Replay-Speed Rule — What the WC2026 VAR Record Actually Supports
An applied study of all 104 World Cup 2026 matches: slow-motion review demonstrably inflated severity in the anchor case — as a breach of VAR's own protocol — but the record does not support a tournament-wide harshness bias, and says so.
On this page6 sections
The question
Is the hypothesis that slow-motion replay biases VAR decisions toward harsher punishment supportable from the compiled WC2026 card and VAR evidence?
That question was pre-registered before the tournament ended, with four rival hypotheses committed in advance — including the two that would kill the founding intuition. The answer below was reached on the complete record: all 104 matches, 282 yellow cards, 16 reds, and a 54-incident VAR catalogue in which every row carries its external source.
Where it started
The study began with one man watching a match. The founding account, preserved verbatim in the study record: in real time the referee did not judge the Balogun contact (USA–Bosnia, round of 32) worth even a yellow; only the slow-motion review produced a straight red. The owner's theory: real-time and slow-motion are epistemically different inputs — a freeze-frame of a boot near an ankle looks like a stamp, while the full-speed sequence reveals a common, non-aggressive play-through between two players challenging for a ball.
The same match supplied its own control. Malik Tillman's foot was cut open by an earlier contact — studs through the boot, bloody sock, widely reported — and drew no card and no review. The contact that drew blood drew no sanction; the contact that drew no visible injury drew a red. (The comparator is deliberately kept out of the incident catalogue: no outlet raised it as an officiating controversy, and the study's selection rule — external markers only, never self-judged — was not bent for a convenient example.)
What the rulebook already says
The sharpest finding in the study is not statistical. The IFAB VAR protocol itself states that slow-motion replays should be used only for facts — point of contact, position — and that normal speed must be used for the intensity of an offence, because slow motion makes nearly every tackle look violent.
The distinction the owner drew from his couch is already written into the laws of the game.
That reframes the charge, and strengthens it. "Serious foul play" is an intensity judgment — precisely the category the protocol says must be assessed at normal speed. ESPN's VAR Review verdict on the Balogun red: incorrect — "accidental contact in a normal challenge; VAR improperly relied on slow motion." A protocol error, reported as such. The critique is therefore internal: VAR failed by its own published standard, not by a standard this study imported. And it is not an anti-VAR finding — the record shows VAR performing well on the factual questions it was designed for. It goes wrong where it applies factual-domain tools to judgment-domain questions, which its own protocol forbids.
What the full record supports — and what it does not
The study's claim board — 90 claims, each tied to its exact source passages — was closed under adversarial review: 58 confirmed, 31 corrected (nearly all updating quarter-final-stage figures to final counts), 1 rejected, none left open. On that board:
Supportable, as directional case evidence: slow-motion severity inflation is real. The anchor case is confirmed by independent adjudication, and the mechanism — replay speed substituting for judgment — is the one the protocol itself anticipates and prohibits.
Not supportable: a tournament-level harshness bias. Three results block it, and the study reports them without softening:
- Under-enforcement is the more frequent VAR error. On the same independent adjudications that confirmed the anchor case, VAR's commoner failure was declining to punish genuine severity — not inflating trivial contact.
- The "enforcement lottery" is dead. The founding intuition included a different-referee-different-justice component. On the completed card data, by-referee variance sits within chance. The real, grounded variance is by stage: the round of 16 spiked to 4.50 yellows per match against a tournament average of 2.71, and second-half stoppage time concentrated 60 yellows and 5 of the 16 reds.
- The catalogue is a floor, not a census. No body publishes VAR decisions as structured data. The 54 catalogued incidents exist because external markers — independent outlets, official statements, formal complaints — flagged them. What that selection rule cannot see, the study does not claim to know.
The abstention is the finding. Where the evidence runs out, the study stops. A tournament-level causal claim about slow-motion bias does not survive this record, and no rhetorical construction was permitted to rescue it.
Method and provenance
Every claim in the study cites its source passage; every review decision is logged and reversible; the ranked differential — favored reading, defeated reading, live rivals — was recorded with calibrated confidence (moderate, 0.45) rather than asserted. The full evidence packet, with every claim printed beside its verbatim passages, is preserved in the study archive. Verification was adversarial by design: independent review passes attempted to rescue every rejection and refute the most consequential confirmations before anything was recorded.
What would sharpen the answer is equally specific: a structured re-coding of all 54 incidents by question type (factual vs judgment) and replay speed relied upon — a pass that requires analytical definitions the owner has reserved to himself — plus a denominator of comparable uncarded contacts, and broadcast review durations that the current sources do not carry.
Why this is not really about football
The study record closes with a generalization the owner logged midway through: enforcement that is severe but rare does not regulate behavior — predictability does. VAR was the clean, low-stakes laboratory where that mechanism could be watched without moral noise. The essay that carries the principle — the enforcement distribution is the rule — is the owner's to write, in his own voice, and is not this page.