The Invisible Ledger of Umpire's Call: 455 Checks, 0.09 Points, and Reviews Bound to a Blockchain

That night the ball found space beside leg stump, and nobody saw how much space exactly. Second ball of the fourteenth over, a flighted length, the batter forward in defence, the ball into the pad. The on-field umpire said not out. Two balls later the fielding side reviewed. On the broadcast screen the ball tracking ran: pitch map clean, impact point clean, wicket projection clean—but the ball was entering the stump zone less than half its width. Umpire's Call appeared. Review gone, decision standing, and the review would not come back to the fielding side for the rest of the innings.

From a flat in Liverpool I did the arithmetic on paper. A cricket ball has a radius of roughly 3.6 centimetres. The protocol says at least half the ball must be projected inside the stumps for the on-field call to change. A band three or four centimetres wide therefore decides where a match goes, how a points table reads, sometimes where a career stops. The frames inside that band never reach the viewer.

## Why Umpire's Call exists at all ​Two thousand and eight, July, India in Sri Lanka: the Decision Review System entered Test cricket. Eighteen years on the technology has moved a long way and the governance has barely moved at all. Ball tracking is a projection model—release point, post-bounce friction, spin, air movement, pitch behaviour. Every model carries an error bar. The ICC decided that inside that error bar, on-field authority stands. That is not a finding of physics. It is an administrative choice. Anyone who thinks technology drew a neutral line between right and wrong is mistaking the tool for the rule. Technology only stated a limit: this is how far I can measure, beyond this I defer to you. Who sets the limit, and whose word survives outside it, is politics already. Test teams get three reviews, ODI and T20I sides get two. Those numbers are published and auditable. The confidence interval, the calibration frames, the operator's keystrokes are not.

## A ledger of 455 checks and one written threshold ​In 2026 I ran an anonymous account logging every Premier League penalty—380 matches, 92 penalties, 14 I graded wrong. Logging is easy; writing down the standard is hard. At Russia 2026 I watched all 64 matches twice and built the ledger: 455 VAR checks, 20 on-field reviews, a record 29 penalties, each intervention scored against a clear-and-obvious threshold I wrote myself. I emailed the forty-page framework to 11 clubs and two refereeing bodies. Four replied; one offered me a junior analyst role. A threshold that is never written down can never be audited. Cricket's problem with Umpire's Call is not that the limit exists. It is that the confidence level inside the limit never reaches the screen.

## Cricket shows more than football and says less ​Football's VAR was attacked for hiding its audio. Cricket did the opposite: ball tracking on screen, third umpire on microphone. What cricket hides is subtler—model version, confidence level, number of calibration frames, the moment an operator nudged an impact point. In process cricket is far more transparent than football; on the uncertainty of its own decisions it is close to silent. A viewer who is not told the margin of error treats uncertainty as zero, and invisible uncertainty hardens quickly into conspiracy.

## 0.09 points per game: the table is a verdict ​June 2026, junior performance analyst at Tranmere Rovers, League One curtailed, relegation decided by unweighted points per game. The maths was brutal in its simplicity: 32 points from 34 matches, 0.94 per game, 0.09 behind AFC Wimbledon. I built the scenario model in four days—unweighted, weighted, a 22-team division. Two of my charts went into the club's appeal. The appeal failed. I did not argue. I logged 92 behind-closed-doors matches instead and found the home win rate had fallen from 45% to 38%. The table is a verdict, not a story, and the people who build the table are rarely named. The same is true of the three-centimetre band. The difference is that the club relegated by 0.09 points has a name in the record; the review that fell short by 3.6 centimetres has none.

## 1.88 millimetres: preparation is the only honest method ​On 1 December 2026, on the data floor in Qatar: Japan 2-1 Spain, Ao Tanaka's 51st-minute winner, and the question of whether Kaoru Mitoma's cutback had crossed the line. Tracking data put the ball 1.88 millimetres inside. I had pre-built the decision tree for that exact scenario three weeks earlier and had two operators red-team it. The check ran two minutes and twelve seconds. Preparation does not shrink the margin. It makes a margin that small survivable. Walk into an international match without that tree and you improvise in front of the world. And the kit itself is unevenly distributed: outside the international circuit, a domestic first-class match in Dhaka or a county second eleven has three cameras, not thirty. Where there are cameras there is evidence; where there are none there is narrative.

## A blockchain question: who owns the ledger ​This is where a ledger design becomes relevant—not as cryptocurrency enthusiasm, but as governance. Every review check written to an append-only chain: hashes of input frames, calibration frames, model version, confidence interval, which keyboard moved the impact point, timestamp. Each entry carrying the hash of the one before it. Two things follow. Nobody can edit the log afterwards, or quietly swap a model version in the past. And anyone can reconstruct a check and compare. The real question about a ledger is never whether it is neutral. It is who owns it. Broadcasters want the controversy; boards want the crisis manageable; vendors want the model proprietary. Three interests, one conclusion: opacity pays.

## The other side: emotion against the rule ​The fan's argument is not weak. Umpire's Call protects wrong decisions in the name of human authority, and technology had already proved the error. My discomfort runs the other way. Remove Umpire's Call and the software becomes the umpire—and then no individual is accountable, which is convenient for everyone. My objection is to the arithmetic, not the authority: where is the document that settled on fifty per cent? Watching football VAR and cricket reviews side by side from Liverpool, one pattern never appears in any dataset. Reviews against big sides get fewer replays than reviews against small sides. The stadium roar fades from memory; the clip that runs from six in the evening until midnight does not. That is not a conspiracy, it is broadcast economics.

## Publish the confidence interval ​My proposal is modest and therefore sounds impossible. After every review, three numbers on screen: tracking confidence, the hash of the configuration used, the duration of the check. Those three show the viewer not the decision but the certainty behind it. When the margin decides who stays, the frame rate is a moral position. I started with a 455-check ledger and I end with a question about the watchers. If the ledger keeps being written only by the people who benefit from it, today's Umpire's Call becomes tomorrow's fifty per cent boundary—thinner by a centimetre, and no easier to argue with.