In a series of new Reddit-posts, Arlington69 claims that the AI positions the players differently depending on the score line. Among the claims is that the defending team’s players apply higher pressure. We of course checked out the validity of the claim.
Arlington69 has been looking into how the AI-controlled players are positioned at different score lines. He posted screen dumps and video clips showing how defenders are positioned differently when the game is tied and when you are leading 2-0. The two screenshots below catch the gist of his claim. At 2-0, we see the red team (trailing) marking the white defenders. At 0-0, we see that the red team keeper has plenty of passing options open. The two screenshots were made with Arlington69 controlling both teams and without making any tactical changes. Clearly, the AI lines up differently in the two situations.
In his summary post, “Why I believe in momentum”, he points to his observations in regards to defensive positioning as one of his reasons to believe in it:
“For this I controlled both player and just stood in one place with a striker and with the goal keeper. I then scored 2 goals and repeated. After scoring 2 goals the player who was loosing the AI behaved differently marking more closely and moving more rapidly. This experiment is repeatable and I believe is solid evidence the game mechanics shift depending on factors within the game. Or there is a momentum effect within the game.”
Arlington69’s methods so far haven’t been highly scientific, and this case is no exception. These are just some of the issues we found when looking into this case:
- Confusing correlation and causality and failing to demonstrate correlation first
- Basing a conclusion on results which aren’t repeatable
- Jumping to conclusions
We are going to do a quick presentation of each issue below.
Santa and the presents
You have two photos – one of Santa with a present and another one of your dad without a present. Does this prove that Santa brings the presents? Probably not. And for the exact same reason, two photos showing the players lined up differently at 2-0 and 0-0 obviously doesn’t prove that the AI lines your players up differently depending on the score line. There is no proof of correlation here, and for that reason absolutely no proof that the players line up differently as part of a deliberate effort to make matches more even.
In fact, Arlington69’s own “data” could suggest that they aren’t correlated. A quick read through Arlington69’s comments to the original series of screenshots reveals the following comments to two screen shots taken when the score line was 2-0:
“At 2 – 0 down no pressure from the AI defence”
“After 5 Minutes still no pressure from AI defence”
(– Comments from Arlington69’s photo section on Imgur)
Apparently, the changes in defensive pressure didn’t happen when the other team went 2-0 up. And yet, Arlington69 concludes that it must have happened because the other team took the lead.
So, if we start out with the basic claim that the AI makes your team behave differently depending on the scoreline, it isn’t justified by the evidence presented by Arlington69. We obviously can’t rule it out either at this point.
Are the results reproducable?
Another problem is that the results aren’t reproducable. In science, it’s a basic principle that anyone with the right skills and resources should be able to reach the same results by repeating the experiment. Arlington69 claims that his observations above are repeatable, but when another redditor tried it out in a slightly different setting, he didn’t see any changes in AI behavior depending on the score line:
“I will get downvoted to hell for going against the grain, i just redid this experiment, with default cts, default instructions, same formation. I played pay a friend squad on ut controlling both teams (different controllers), and did series of test, 1-0 1-1 , 2-1 2-2 3-2, at different points. and the ai reacted very similarly.”
(– Comment on Reddit)
If Arlington69 decided to test his assumptions under different circumstances, it’s possible that he would see teams leading lining up more aggressively as well.
Skipping the stuff in between
What we seen in the pictures above is a team lining up differently in two situations. It may make a difference how teams line up, but even if we had established that there was a connection between score line and line up, that wouldn’t lead to the conclusion that momentum exists. In order to establish that, we would need evidence supporting that the way players line up has a significant impact on the end result. We don’t have such evidence, so for all we know, teams lining up differently in some situations could be something as trivial as the game making matches varied.
Does the score line decide the defensive pressure?
Arlington69’s posts are not scientific, and his conclusions are not supported by his data. The case we have investigated here is no exception. The evidence he has provided does not prove that the AI increases the defensive pressure for the losing side. And even if we were to establish such a connection, that still wouldn’t lead us to conclude that momentum exists.
If the game was leveling matches by altering the defensive pressure depending on the score line, this would reflect in the result statistics. But when we look at the result statistics, we see that matches are less even than real football matches.