The Three-Judge Lottery: The Arithmetic Hidden in Paris's Scorecards

3 August 2026, North Paris Arena. The 71kg quarterfinal ends and the screen reads 1-4. Nishant Dev walks out of the ring. I was watching from London, a notebook open in front of me with the same question I write at every major tournament: which of the five judges gave which round to whom? That answer never appeared on any screen that night. Boxing announces the verdict. It does not announce the arithmetic.

In football that same week, a penalty-area handball had already produced four frames, two camera angles and an audio clip in the public domain. I had spent the 2026 World Cup timestamping every VAR review with a stopwatch and a spreadsheet — 64 matches, 20 changed decisions. The habit that stuck was this: count the reviews first, then listen for the silence beside them. In Paris, the silence beside the boxing was very loud.

Context: what the rulebook says

The Olympic format runs on the 10-point must system. Five judges per bout. Each judge scores each round 10-9, or 10-10 for an even round. And here comes the clause spectators rarely know: the round scores do not simply aggregate across all five cards. For every round, three of the five cards are selected at random, and the majority of those three decides that round.

The lottery is not evidence of a plot. It was designed deliberately, so that one judge's bias could not write an entire bout across three rounds. But every design has a price, and the price of this one is variance.

The governance timeline matters here, because who wrote the Paris scorecard is not a question that rests with one body. The Indian federation was suspended in 2026; the BFI was formed in 2026. AIBA renamed itself IBA in 2026. On 22 June 2026 the IOC withdrew IBA's recognition. Paris 2026 was run not by IBA but by the IOC's Paris 2026 Boxing Unit. On paper the rules were the same; behind the desk, the institution was different. World Boxing was founded in April 2026, received provisional IOC recognition in February 2026, and boxing was added to the Los Angeles 2028 programme in October 2026.

Core: where a verdict hides its arithmetic

In March 2026 boxing stopped, Tokyo postponed, and I had eleven weeks. In those eleven weeks I built 'The Card' — 1,340 Olympic-format bouts from London 2026 through the 2026 World Championships, every judge's every round, transcribed from IBA bout sheets. Eleven weeks of scorecards showed me the place where a verdict hides its arithmetic.

The clearest finding: in 38% of bouts decided 3–2, at least one judge awarded rounds to the opponent that the other four awarded unanimously to the other boxer. Roughly one split in three carried a card in direct conflict with four-judge consensus — and that card held a formal right to flip a round, if the draw selected it.

The mathematics here is simple, and it is the core of what I found. If two of five judges give a round to boxer A and three give it to boxer B, there are 10 possible combinations of three cards. Only 3 of those combinations include both minority judges. Which means a 2-of-5 minority wins that round roughly 30% of the time. A lone dissenting judge can never win a round — his voice is arithmetically inert. Add a second and that voice comes alive at thirty per cent.

Apply this to a three-round bout. If a 2-of-5 minority forms in any one round, the scorecard records that round as a majority verdict — and anyone reading the bout sheet later has no way of knowing whether two of five won it or three of five did. In the card, both look identical.

The second pattern circles the host nation. In bouts involving a host-country boxer, round-swing variance ran above my dataset average — and the effect was strongest in the early rounds, precisely when the pressure to bank points is highest. That is not proof of corruption. It is crowd and atmosphere, but it is an effect, and it can be measured.

The third result is the least interesting and the most useful: in bouts where all five judges agreed on every round, the outcome is formally as valid as any split. The system files both the same way. To a reader they are very different things; to the record, they are one thing.

Contrarian: publishing all five cards is not the fix

The most popular demand after Paris was to publish all five judges' cards immediately. I do not oppose it, but I think it is incomplete, and I think it misses the actual gap.

The gap is review, not publication. The football comparison works here only as a control case, and only up to a point: football's review functions because its standard is binary — an offside line either is or is not. Boxing's standard is not binary. 'Clean scoring punches' is a judgement, not a measurement. Bolting a VAR-style door onto it is decorative. What can be built is an appeal on arithmetic inconsistency — if one judge awarded rounds that the other four unanimously reversed, that card should face a defined review, with a written outcome.

The protest window that exists today closes 30 minutes after the verdict is announced, and its grounds are limited to procedural error; a judge's evaluation of the bout cannot be challenged inside it. The system has closed itself to the scoring question and placed a fee-collection box at the shut door. Of the bouts in my dataset that drew the loudest public objection, how many produced a formal protest was never recorded the way the result was recorded. I counted the reviews first, because the silence was harder to count.

Takeaway

World Boxing's LA28 qualification system is due in 2026, and that same document will decide how judges are appointed, when cards are published, and on what grounds a protest is heard. The question is not whether boxing will keep producing disputes — it will, because a human decides inside the ring. The question is what the 2026 paperwork will say: 'the result was announced', or 'the result, its arithmetic, and the route to appeal against that arithmetic'?