The 3.6-Centimetre Umpire's Call: A Margin Wider Than the Stump

The last frame of ball-tracking froze on the big screen. The ball was grazing the outer edge of leg stump — two millimetres of it, maybe three. Two rows behind me a man was on his feet shouting, because on the replay it plainly looked like ball meeting stump. Then the line everyone in that ground has learned by heart appeared: the ball is clipping, the on-field decision stands. Out. I watched that match with a notebook and a pen. That is the old habit — not a scorecard, a ledger of decisions. In that single moment the crowd and the umpire are not watching the same event, and both of them are right. A decade ago, tagging highlight clips at a Dhaka production house, I built a taxonomy of 412 clips. It collapsed twice and had to be rebuilt, because the first version broke on the dismissed-handball category. But when I opened that ledger I found a pattern nobody had named: the argument is always about the replay, and the decision is always about a margin. That margin is 3.6 centimetres. DRS first walked onto a field in July 2026, in the India–Sri Lanka Test series. The purpose then was narrow — protect the umpire's reputation, not install technology as the final judge. In June 2026 the ICC changed the LBW protocol, and that is where the story turned complicated. Under the amended rule, if ball-tracking shows the ball is not hitting the stumps fully but merely clipping them, the call is umpire's call — the on-field decision survives. Which raises the only question that matters: how much is merely clipping? A cricket ball measures roughly 7.2 centimetres across. Half of that is 3.6 centimetres. A single stump is 3.49 centimetres wide. So the tolerance zone inside which the decision returns to the umpire is wider than a stump. Most viewers assume umpire's call means the ball brushing the line. It is actually a corridor larger than the target itself. I call it the invisible budget, because the thing the ground cannot see as data is exactly what the wicket is accounting for. An LBW review is three separate layers — where the ball pitched, where the impact occurred, and where the ball was travelling. Each has its own tolerance and its own decision gate. Umpire's call almost never appears at the pitching layer; nearly all the traffic sits at impact and at the stumps. Teams that treat those three layers as one question usually burn the review, because they are not reading evidence — they are reading a thick television replay. There is another number in the protocol almost nobody discusses: three metres. If the point of impact on the pad is more than three metres from the stumps, the on-field umpire's decision is no longer automatically protected; the ball-tracking projection becomes final. The further a batter advances, the less human tolerance he is granted. Read the two rules together — strict on distance, soft at the edge — and the design reveals what it truly values: the position of the batsman, not the flight of the ball. Batters know this corridor exists. It is why so many of them stay unconcerned after blocking a ball with a leading pad; they know that even if part of the ball is touching the stump, the decision goes back to the ground. For spinners the arithmetic gets worse. Once the ball lands in the rough and changes direction, the projection carries more doubt, and clipping calls multiply. The folk belief is that spinners suffer most from umpire's call. The ledger says the variable is not the bowler but the behaviour of the pitch. Reviews are a finite resource — two per innings in Tests, one per innings in ODIs and T20Is. Since 2026, Test sides have their reviews topped back up after 80 overs, which quietly favours teams that bowl long and stay patient. For the last few seasons my own ledger has split reviews into three buckets: the protest review, the plan review and the panic review. The protest and the panic are usually gone by the tenth over. When the last hour arrives and a review actually matters, the hand is empty. The real argument, then, is not about the rule. It is about habit. No side reviews out of a sense of injustice — they review under pressure, at a bowler's pleading, sometimes on a captain's own uncertainty. The nightly ledger did not lie; it just waited for me to read it again. Read again, and the sides that bank a review for the last ten overs are the sides with more chances to intervene in the two decisions that decide the match. That is a correlation, not a cause, and I mark it as such. Ball-tracking tolerance is a fixed number everywhere, but projection error is not. On slow, low-bounce surfaces the ball deviates more than once before reaching the stumps, so the same 3.6-centimetre corridor carries far less confidence. On a hard, true-bouncing pitch the path is easier to model and the corridor effectively narrows. That is my hypothesis, not a proven citation — but the shape of the ledger points that way, and where the data runs out I say so before someone else does. In 2026 the Bangladesh Premier League was abandoned, the stadiums emptied, and I moved down a layer into sound. In closed-door footage I was listening not to crowd noise but to stump-mic murmur, players' whispers, the umpire's reluctant half-sentence. Cricket already treats sound as evidence through UltraEdge, and yet we have never learned to read audio as a witness to decisions. When the stadium emptied, the microphones started testifying. This is where my sharpest objection sits. The same ball-tracking data that settles a wicket is piped into live markets within seconds, and the same 3.6 centimetres that flips a decision also flips a price. The darkest consequence of the sport's datafication is right here: the line that decides a match also decides someone's income. And still my ruling is that umpire's call is not a technological failure. It is an institutional design whose job is not to interrogate technology but to preserve the authority of the on-field umpire. The cry of being robbed belongs to the wrong category. Nobody is cheating; the system is behaving exactly as it was built to behave. The more uncomfortable problem is statistical. In the familiar claim that 94 per cent of decisions are correct, umpire's call counts as correct by definition, because it is the on-field decision. The measuring stick is grading itself. The crowd tells you what happened; the tape tells you what actually occurred — but the last frame of that tape always conceals a human reservation clause. I have not stopped trusting my eye. I built a calibration protocol instead: not the beauty of the replay, but the height of the ball, the bounce of the pitch and the width of the corridor — those three, in that order, are how I read a review. Technology will take another step. I hope the next change arrives as transparency. Publish venue-level confidence around the projection, and both teams and crowds will finally understand when a review is pointless. I have written these forecasts into my own ledger and will audit them at the end of the season. The question stays open: when the tolerance is wider than the stump itself, is the game measuring the ball — or the umpire?