Half a Ball from Umpire's Call: When DRS Predicts Instead of Measuring

Hook

The scene repeats itself almost identically. A spinner releases, the batter is pinned on the pad, the on-field umpire shakes his head — not out. The batting side reviews. On the giant screen the yellow ball appears, clips the inside edge of leg stump and travels on, and two words flash up: Umpire's Call. The stadium roars. The roar is aimed at whom, exactly? The umpire? The technology? Or that yellow ball, which never actually rolled anywhere — a model merely calculated that if it had rolled, it would have rolled like this.

I have watched that moment frame by frame more times than I can count. In May 2026 the ICC Chief Executives' Committee approved the abolition of the soft signal, effective from June that year. In catch reviews, the on-field umpire's provisional verdict no longer exists. On paper this was a victory for evidence — no third witness standing between an umpire's guess and the tracking model's arithmetic. On the field, the result inverted. The soft signal vanished; the anchor did not. It simply changed address, from the umpire's head to the tracking model.

Context: The same text, read differently by everyone

A misconception clings to DRS: that it is a truth machine. It began in 2026 as an experiment, in Sri Lanka during India's tour, giving players the right to request a review of a decision. Early on, every board, broadcaster and series applied its own rules. After 2026 the structure settled: two unsuccessful reviews per innings, a review reset after 80 overs in Tests, a successful review returned.

The system rests on three pillars — ball tracking, audio edge detection, and once upon a time Hot Spot. Into that was bolted the soft signal: on catch reviews the on-field umpire would state a provisional call, and the third umpire would anchor the final verdict to it. In 2026 came the three-metre rule — if the point of impact is more than three metres from the stumps, the umpire's-call concession on impact-in-line no longer applies, and the third umpire checks only whether the ball would have hit. From June 2026 the soft signal was gone.

I watch this system from Singapore, at an odd distance. Matches start late at night here, the feed passes through broadcasters' hands, and the decisive frame arrives about thirty seconds after it was decisive. That delay is not innocent. The review system sells three things at once: a measuring stick, a process, and a settlement of power. The first is mathematics, the second protocol, the third politics. Cricket arguments are almost always about the first, while verdicts are decided by the third.

Core: Inside the tracking room

One: what is measured, what is forecast

Ball tracking does not watch the ball the whole way. The stretch before pitching and the first few feet after are captured in frames; once the ball hits pad or bat and stops, the remaining path is computed by a physics model — spin, bounce, air, seam, grass. A large part of the yellow ball that enters the stumps on screen is not observation but prediction. In reporting language: that is not testimony, it is expert opinion — and expert opinion always carries an error margin.

Why does this distinction matter? Because when a measurement errs, an apology is possible. When a forecast errs, nothing can be done except to rewrite the explanation. The moral foundation of the review system stands on that uncomfortable spot, and the screen never says so. It shows only a yellow ball and a white batter.

Two: the geometry of half a ball

There is a number in the ICC playing conditions that is cricket's least-discussed number: half the ball. Only if tracking shows more than half the ball entering the stumps does the on-field decision change. Half or less, and the decision stands — that is Umpire's Call. What happens here is a continuous probability abruptly dropped into a binary. The model may be saying there is a forty per cent chance the ball hits. The spectator saw a yellow ball graze the stumps. Forty per cent is not five per cent to them, nor zero — it is an image, and an image never becomes a probability.

That translation loss is the roar in the stadium. The roar is not against the model; it is against the translator. The ICC publishes a probability as a final word, then seems surprised that people do not believe the word.

Three: simulating the umpire's eye

Before getting angry, one task is due — stand where the umpire stands. From square leg, a spinner releases at roughly twenty metres per second, the ball pitches and breaks six to ten degrees. Impact lands on the inner edge of the pad, and the front leg hides the last stretch of the ball's travel. What the eye captures is pitch, line, height. What it cannot capture is whether the ball would have entered the stumps. And that missing information is precisely what tracking later claims to supply. The umpire's eye is not a camera; it is a compromise under pressure — and the law has conceded that this compromise is indispensable.

The three-metre rule matters here. When a batter steps out, the distance between impact point and stumps grows, and so does the model's uncertainty. The rule therefore removes the umpire's-call concession on impact-in-line. The law itself admits: beyond a certain distance our model is not good enough, so the question is dropped. Such self-confession is rare in cricket's statutes, and it is the most honest part of them.

Four: the third umpire's protocol and feed latency

Verdicts arrive in the third umpire's room in a fixed order: front foot for the no-ball, then audio for bat contact, then pad-ball connection, then impact-in-line, finally tracking. That order determines which evidence speaks first, and reordering it can reorder the verdict.

Here the feed-first question enters. On broadcast we see a settled sequence, sometimes in slow motion, sometimes from a side angle, sometimes with graphics laid over it. The decision, though, was made in a different order, on a different screen, at a different frame rate. I follow the audio, the angle and the agenda — usually in that order. If the stump mic catches no bat click while UltraEdge shows a spike, two pieces of evidence stand against each other. Who wins that contest is not a question of technology but of protocol: who testifies first, and who decides that.

Five: the ghost of the soft signal

The abolition should have been a clean release. No more sitting in the shadow of the field umpire's guess. But an old psychological finding reaches out: anchoring. In 2026 two researchers showed that when people encounter a relevant number before evaluating something, their final judgement is dragged toward it. The soft signal was a textbook anchor, because it was an informal verdict. Remove the anchor and what happens? The judge turns his own first impression into the anchor.

In tracking, that new anchor is the model's first output. Before the third umpire even sees the image, an expectation has formed — how far forward the batter was, how much the ball turned, what the umpire said. Call it a background anchor. Where the soft signal was loud and declared, this one is silent and deniable.

Six: doors into other games

Rulebooks are adjacent rooms with unlocked doors. Football's VAR arrived at the 2026 World Cup, and in areas like handball the gap between written text and application became the biggest political issue of all. I slowed the handball down until intent and geometry stopped agreeing. Semi-automated offside arrived at the 2026 World Cup, removing the human call by drawing a line. Tennis adopted Hawk-Eye Live long ago, because a line is a geometric question, not a probabilistic one. Cricket's DRS chose the opposite road: instead of measuring, it kept the probability — and handed ownership of that probability back to the umpire.

Personal experience is not irrelevant here. Since moving from Bangladesh to Singapore, one thing is clear: the same written rule means different things in two jurisdictions, because administrative hands and public expectations differ. Cricket's threshold rules work the same way — identical ISO-style numbers, but never neutral to the person they are applied against.

Seven: my own logs

At the 2026 World Cup I logged all 29 penalties and 17 VAR-reviewed incidents, and that was where I first understood that technology does not remove error, it redistributes blame. In 2026, when the Singapore Premier League returned behind closed doors, I tracked 32 matches and found yellow cards for dissent had risen about eighteen per cent against 2026. Remove crowd noise and referee-player dialogue lands on open microphones, turning every tackle into a disciplinary test case. In an empty stadium, the dissent audio becomes the loudest witness. The lesson transfers directly to cricket: reaction against Umpire's Call is often about how a verdict is announced, not about the verdict.

Across eighteen years of reporting I have watched top-order batters — Virat Kohli, Steve Smith, Kane Williamson, Babar Azam — use reviews not only as instruments of justice but as innings-management weapons. Experienced players like Shakib Al Hasan and Mushfiqur Rahim know which umpire tends to rule which way, and that knowledge shapes the next review strategy. This is where law and tactics swim together.

Contrarian: the anger is aimed at the wrong address

Umpire's Call is not a defect; it is a deliberate subsidy. The law has kept a slice of its authority with the on-field umpire, and removing that subsidy would make the field umpire decorative in the literal sense — not deciding, merely waiting for the tracking model. Does cricket want that? The evidence suggests not. Spectators do not want technological infallibility; they want a tolerable compromise between technology and custom.

So where is the real problem? Increasingly I believe it is not in the threshold but in the announcement format. The law translates a continuous probability into a binary word on screen — Call or Out. If the magnitude of tracking uncertainty were printed alongside, Umpire's Call would not be a synonym for insult. A second problem: as technology grows, so does the number of thresholds — fifteen per cent, fifty per cent, three metres, one frame. Rules are fixed text, but enforcement is a living, sweating thing. Each new threshold moves authority from the field to the studio, and we call that progress.

The materiality test matters here. How many of these arguments actually change outcomes? Very few. But even when outcomes do not change, the balance of authority does — and that balance defines the character of the game. That is the real event, the real story.

Takeaway: who writes the next threshold

Semi-automated offside has shown the road. Cricket's next big threshold will come in no-balls, then edge detection, and eventually perhaps at the stumps themselves — where the model rules and the umpire stands as witness to the process. By then the question will no longer be whether Umpire's Call survives. It will be: who trained the model, on whose data, and who announces its uncertainty? When forty per cent becomes fifty-one per cent, who is the umpire — the human, or the training set?

Sources: ICC Chief Executives' Committee approval in May 2026 and abolition of the soft signal from 1 June 2026; the three-metre rule introduced in 2026; the author's own match logs (29 penalties and 17 VAR incidents at the 2026 football World Cup; 32 matches of the 2026 Singapore Premier League) — cross-checked: cricsultan.com