If you are a football or soccer fan, then you may be somewhat informed about VAR. VAR is essentially a replay centre, and that in itself is not technologically advanced in any way. In fact, the use of VAR is probably somewhat outdated already, and should have been around twenty to thirty years ago. It had been introduced way too late in the game of football and soccer, and it seems only natural that the upgraded and better version of VAR is already in preparation.
This specific project that we decided to invest into is not simply a replay room. The problem with VAR in its current format as we have seen in the Women’s World Cup 2019 is the ability to still make the incorrect decisions regarding refereeing decisions. Even worse, it has taken a total of 45 minutes of in game playing time for those decisions to be made. What we have on our hands is not a replay centre, but rather an actual VAR referee.
The VAR referee is simply an algorithm based on hundreds of thousands of football footage. Using the same techniques as you would a machine learning AI, a human had sifted through hundreds of thousands of hours of fouls and other bad decisions by players to teach an AI script on what decisions are incorrect and when a foul or a mistake has been made.
Using this same algorithm, the VAR is now able to pick up a foul or a non-foul correctly 97% of the time instantly and will automatically ping the referee with a code number whenever a foul is made, or a decision needs to be reversed.
However, due to the infancy of the project there are still mistakes being made on a daily basis with the 3%. Until that 3% is ironed out and there is a 100% success rate, there is no reason to implement the code. There will also be a margin of error for new issues that may arise, but the group cannot plan for that.