How Does Video Review Work in Sport? Replays, Thresholds and Arguments
By the BrainSnail editorial team. How these articles are written and checked, and how to tell us when one is wrong.
Every major sport now has some way of checking a decision on video, and every one of them has found that removing obvious errors does not remove controversy. The reason is that the technology answers a narrower question than spectators assume: it does not ask what the correct decision was, it asks whether the original decision was wrong enough, by a specified standard, on the specified categories of incident, within the specified time.
The two designs
Systems divide into challenge-based and official-initiated, and the choice shapes everything else. In a challenge system, used in tennis, cricket and American football, a team or player has a limited number of requests, typically two or three, and usually retains them if successful, which forces them to spend a scarce resource on their own judgement of how wrong the call was. In an official-initiated system, used in football and rugby, a separate video official monitors play continuously and intervenes on specified categories without anyone asking. The first creates strategic decisions and leaves genuine errors uncorrected when challenges have been spent; the second removes that arbitrariness and creates a different problem, since every incident is now potentially under review and play can be recalled after it has moved on.
The threshold
The most misunderstood element is the standard for overturning. Most systems do not ask the video official to make the decision afresh; they ask whether the evidence shows a clear and obvious error, so a marginal incident on which two reasonable officials could differ stands as called. The design reason is that a system which re-decides everything makes the on-field official redundant and slows the game to a halt, and the practical consequence is the exchange that frustrates viewers, in which a replay appears to show one thing and the original decision is upheld because it was not wrong enough. Cricket formalises this most explicitly with its umpire's call provision on leg before wicket dismissals: where ball-tracking shows only a marginal portion of the ball hitting the stumps, the on-field decision stands whichever way it was given, so two identical deliveries can produce opposite outcomes depending on what the umpire said first.
What the technology can and cannot measure
Some questions are genuinely binary and machine-answerable, and others are irreducibly matters of judgement:
- •Goal-line technology, which uses multiple high-speed cameras to determine whether the whole ball crossed the line, is accurate to millimetres and reports to the referee's watch within a second. It is uncontroversial because the question has an exact answer
- •Ball-tracking in tennis and cricket predicts a trajectory from camera data, with a published margin of error of a few millimetres, and its cricket application involves predicting a path the ball never took
- •Semi-automated offside systems use limb-tracking and a chip in the ball to determine the moment of the pass and the relative positions, which reduced decision times substantially and does not settle what counts as a deliberate play or an active interference
- •Foul severity, intent, handball, dangerous play and obstruction are judgement calls that no camera resolves, and these are where nearly all the disputes now sit
- •Timing of an event, which is what offside depends on, is affected by frame rate, since a camera sampling at fifty frames per second cannot locate a moment more precisely than twenty milliseconds
What it did to the experience
The costs are real and were mostly unanticipated. Celebration is the clearest: a crowd that has learned a goal may be recalled two minutes later hesitates before celebrating, which changes the emotional shape of the event that people pay to attend. Delays accumulate even where each is short. Communication has been a persistent weakness, since in several sports the crowd in the stadium cannot hear the review or see what the officials are looking at, a gap that the sports which broadcast the referee's explanation, including rugby union and American football, handle far better. And accountability shifted rather than increased, since the on-field official can now defer, and the video official is usually anonymous. Several studies have found that video review reduces clear errors substantially, which is the thing it was built to do, and has not improved perceptions of fairness, which is the thing people thought they wanted.
Where it is going
The trend is towards automating what can be automated and giving up on the rest. Fully automated line calling replaced human line judges at several tennis tournaments and across the whole ATP tour from 2025, removing an entire category of official and, with it, the ritual of the challenge. Semi-automated offside is now standard in major football competitions. Sensors in balls, pucks and equipment are increasingly common. What is not automatable is the judgement layer, and the live debate in most sports is about how much of that layer to expose: whether to publish audio, whether to explain decisions publicly afterwards, whether to allow coaches to challenge subjective calls, and whether the threshold for intervention should be lowered so that the video official decides more, which would trade delay and on-field authority for accuracy. No sport has found an arrangement that satisfies everybody, and the underlying reason is that a rulebook written for human judgement cannot be adjudicated by a camera without someone deciding what the rules actually mean.
The takeaway
Video systems are either challenge-based, spending a limited resource on a team's own judgement, or official-initiated, with a separate official watching continuously, and most overturn a decision only on clear and obvious error rather than re-deciding it, which is why marginal replays leave the original call standing. Technology answers exact questions such as whether a ball crossed a line, and cannot answer intent, severity or handball. The result is fewer clear errors, delayed celebrations and no improvement in perceived fairness.