Issues in Interdisciplinarity 2019-20/The evolution of evidence in football

Football
The sport of football, is a game where 11 men or women face other respective 11, typically played on a green turf (artificial or natural) measuring 100-110 meters by 64-75 meters, where the objective of the game is to get the ball into the goal of the opponents mostly using the feet. Each team organise themselves in 4 positions (goalkeeper, defenders, midfielders, and attackers) and try to combine among themselves in order to score a goal. A match normally lasts 90 minutes, divided into 2 halves of 45 minutes. Along with the 22 men or women competing against each other, there are also typically 4 referees that are employed to enforce the laws of the game. What happens when a law is broken? How do we know the law has been broken? Can we be sure the law has been broken? are questions that referees typically ask themselves during the game: their task is therefore to use an available body of evidence in order to make a decision in relation to a play. For example, wondering if there was or wasn't a handball, or if a certain action should lead to a penalty or if it shouldn't.

Evidence
We will take a look at the evolution of evidence in football as it is tightly linked to the role of referees. According to the Oxford Dictionary of English, evidence is "the available body of facts or information indicating whether a belief or proposition is true or valid". These available body of facts/information are sources of evidence, and they can be of different types : quantitative or qualitative. Quantitative evidence relies heavily on numbers and mathematical observations whilst qualitative evidence is descriptive and conceptual. In football, both types of evidences are present. The role of evidence is to support and possibly prove a certain claim

Bayesian Evidence in Traditional Football
In a football game, until very very recently, referees were limited technologically, and could only base their decisions based on their immediate interpretation of the action plus their prior knowledge. For example, to determine whether there was a penalty or not for a specific action, referees could only use their own, immediate, biased observations as evidence. This very often led referees to a Bayesian way of dealing with situations: that evidence should be used to update prior knowledge. To illustrate how Bayesian thinking normally influences the decisions of referees, we will take a look at the player Neymar da Silva Santos Júnior, also known as Neymar. Often considered one of the best modern football players, Neymar is very often the target of many fouls perpetrated by the opposing team, as the only way to stop him from scoring goals is to lightly injure him, or stop him in his tracks. During the 2018 FIFA World Cup Game Brazil vs Switzerland, Neymar was fouled a record number of 10 times. This very high number of fouls on Neymar led referees to call out fouls on him much easier, because they knew he was a very high profile target. The probability that the referee would call out the 10th foul on Neymar during that game is much higher than the probability the referee would call out that exact same foul on some other player. Why is this? Because the referee was influenced by the high recurrence on fouls on Neymar, and was thus very likely to call out that 10th foul. Were a journalist to interview this referee, the actual evidence he would provide to support his decision would be very subjective: "I saw what I saw, and it was clearly a foul", and this is because the referee has supreme, irrevocable, power of decision.

The evolution of bayesian based to frequentist based evidence in football with the introduction of VAR
The actual evidence used in football to support a decision has taken a dramatic twist in the last couple of years with the introduction of the Video Assistant Referee, or VAR. This technological innovation has allowed an extra match official to review decisive actions in a game using video replays of the action in question. This new innovation requires the installation of cameras all over the stadium in order to get a maximum quantity of different angles for this extra referee to review the action. This means that when a decisive play occurs, referees will finally be able to provide concrete evidence of their decision. On the much-evoqued 10th foul on Neymar during the aforementioned game, the referee can use VAR, review the action using multiple angles, and concretely conclude that it was indeed a foul. The evidence is still partly Bayesian, as the referee will be influenced by the fact that this is the 10th time Neymar is fouled, but statistically speaking, the probability that the referee calls the foul will resemble that of Frequentist Probabilities because the referee will be able to rely more heavily on actual, independent to the other fouls, evidence, as the actual footage of the foul will be objective.