The Third Umpire's Clock: How Two Minutes of DRS Is Draining the Rhythm of a Match

Last January at the Chinnaswamy, I was timestamping something the broadcast cameras never show. Seventeenth over, third ball of a right-arm quick's spell. The on-field umpire gave it out, the batter reviewed, the third umpire ran ball-tracking. From appeal to final verdict: 3 minutes 41 seconds. The review held, the wicket fell. Then the seamer walked back to his mark, rolled his arm twice, and his first ball was a half-volley — four.

In the margin of my notebook for that evening I have written: in the over following every review that took longer than 2 minutes 30 seconds, he bowled four half-volleys in total. Four is not a sample, and I know it. But it made me open five years of DRS ledgers.

Context: what the system is, and what I can actually see

DRS first went live in 2026, in the India–Sri Lanka series, when ball-tracking and real-time snicko were both experimental add-ons. Today the third umpire works with three separate engines: Hawk-Eye's ball-tracking, which models pitch, impact and stump height; UltraEdge; and multiple camera angles. Under the ICC Men's ODI Playing Conditions, the on-field side gets only 15 seconds to decide on an appeal — that is the system's only hard time limit. There is no written second-cap on the third umpire's own process. So the length of a review is set by technological complexity and the number of angles, not by anyone's haste. When the ICC introduced two reviews per innings in T20 internationals in October 2026, the question of length became more relevant, because a larger quota also enlarges the potential for waste.

My argument rests on a dataset I built myself. Between 2026 and 2026 I logged 412 reviews across 118 men's internationals — Tests, ODIs and T20Is combined. Each row has four columns: timestamp of the appeal, timestamp of the final decision, the type of decision, and the runs conceded in the six balls after the review. This dataset cannot see player fitness, field placement, or dressing-room pressure. It is only a ledger of a clock and a scoreboard, and it is better to admit that limitation up front.

One regular-season reality matters here. In the relentless franchise and bilateral calendar, third umpires handle two or three matches a day, and bowler workload management is already hard to explain. The quota arithmetic, the over-rate fine and the fast bowler's load are three hands on the same clock, and this season they are landing together.

Core: what the clock says

The first number is simple. Across the 412 reviews, the average time was 2 minutes 38 seconds, the median 2 minutes 21 seconds, and the longest 4 minutes 12 seconds. But the average is the least useful number here, because reviews are not one thing. Split them and the picture changes: LBW or ball-tracking reviews average 3 minutes 4 seconds, while caught-behind or UltraEdge reviews average only 1 minute 52 seconds.

The difference is technological, not human. Ball-tracking has to reconstruct the full trajectory from pitch to stumps in three dimensions, then verify whether the centre of the ball is hitting the stumps, where the pad was in the impact zone, and how much of the pitch is predictive. UltraEdge is far simpler: there is a spike or there is not — a two-second verdict. And yet the two review types sit 72 seconds apart on average, and those 72 seconds are the most expensive real estate in the match.

The second question was my real curiosity: does the delay actually hurt the bowler? My first cut said yes. In overs that began after a review lasting longer than 2 minutes 30 seconds, the bowling side conceded 5.9 runs per over. The baseline for the rest of the tournament was 4.6. A gap of 1.3 runs per over, on a sample of 118 overs.

Because most of that 1.3 is selection effect, and admitting so is my job. When are reviews taken? When the match is hot, when the scoring rate is already high, when the wicket matters most. I was reading a relationship between duration and outcome when both were being produced by the same cause: match pressure. After adjusting for phase of play, the scoring rate of the previous three overs and the state of the wicket, the post-review excess fell to 0.7 runs and the confidence interval touched zero. The raw number is not a lie; it is a raw number, and raw numbers betray their authors.

One thing did survive the ledger, and it was not runs — it was over rate. A three-minute review takes exactly three minutes out of the innings. Four reviews in an innings means twelve minutes. Add rain threat, floodlights and dew to an ODI schedule and those twelve minutes often create pressure at the back end, which returns as a slow-over-rate fine on the fielding side. The time the third umpire takes is billed to the bowling coach.

The third layer is decision quality. Of the 412 reviews, 137 were overturned — 33.3 percent. One wasted application for every three reviews. Teams burn a two-review quota even in situations they call certain, because confidence and evidence are not the same thing. My log shows another pattern: in T20Is the decision to review comes much faster, but the cost is higher, because a burnt review means one hand tied for the rest of the innings. In Tests it is the opposite — five days devalue time, so the review arithmetic is looser.

My viewing experience says there is one specific error behind that 33 percent: the wicket-to-wicket assumption. I have seen many times that on a slow pitch, the batter trusts the model's late movement far less than the model does. So the LBW appeal is often a matter of belief, not proof. The broadcast camera never shows the legal frame; my ledger shows only that.

I looked across at my football VAR logs too, because the method I learned at the 2026 World Cup does not survive transplantation here without showing its cracks. In football a VAR check averages under a minute, and after semi-automated offside arrived it fell further. Cricket's ball-tracking is the same logic of work, but cricket still stops line calls halfway under the label of umpire's call. That is not the preservation of accuracy; it is a decision to preserve ambiguity.

Contrarian angle: where my own model bites me

Now the part where my own model bites back.

Long reviews destroy a match's rhythm — I have written that sentence many times, and it is comfortable because it is hard to test. But when I look at my own ledger, an uncomfortable fact surfaces: of the overs where the run rate spiked, roughly three-quarters had injury, cramp, or a death specialist's workload behind them. The review delay was a spectator there, not an actor. Rhythm breaks on uncertainty, not on duration. Drinks breaks, long changeovers, injury time-outs — cricket has many legitimate reasons to stop, and nobody writes against those.

The second problem, which I learned from the underside of my own dataset: this log cannot see field placement, a bowler's fitness, or the conversation between captain and bowler. Whether the seamer bowling half-volleys after a review has cramp in his calf, the scoreboard cannot say; whether it is Bumrah or Cummins, the fatigue of the workload does not register in this arithmetic. The eye test is a hypothesis, not a verdict — and a coach who forgets that line will push workload management in the wrong direction.

Still, one claim I will defend inside the limits of this discussion: without any control on the length of a decision, the system closes the door on itself. Football's lesson between 2026 and 2026 is the same — technology cannot stop everything just because it is allowed to. In cricket, ball-tracking plots are shown once, then again, then again. The viewer's uncertainty becomes entertainment, and the game becomes a television promo. Football got that down to 90 seconds — not by changing the rule alone, but by changing the number of cameras and the order of decisions.

Takeaway

Two columns stay open for me over the next few series. The first: whether ball-tracking moves to automated line calls. If any series brings the third umpire's average down from 2 minutes 38 seconds to 90 seconds, that is the story of the season for me — however big the scoreboard gets. The second: whether the sides that hand the review decision to an analyst's desk rather than an umpire's instinct can push the overturn rate below 33 percent. A review quota can be traded between formats; the quality of the decision cannot. The column for what the broadcast never shows stays open.