This might be a bit controversial, but I'm thinking the AI might be on a bit of an "elastic band". What do I mean by elastic band? The AI seems to be purposely made to keep in contact with each other for close racing. Why do I think this, I'll explain why. I ran a 20 car Stock Car 2020 race at Goiania. I purposely retired to the pitlane and put the game in monitor mode and watched the AI qualify and race. Settings were AI 97% and agression 50%. The AI qualified at 1m 29.0 on pole and the slowest was 0.6 seconds slower. No problems there it's a very competitive series and the cars are very even. However running the 1 hour race as I watched it all the cars were very bunched up throughout the race. For the first 30 minutes before any pitstop the car running up to 17th place were within 9 seconds of the leader for the entire time. Only varying at most by half a second or so. Realistically cars running and dicing with each other would normally spread out over time, but the the pack remained very compressed. Also in the first lap there was a collision between 2 cars and they went off the track and were quite a few seconds behind. It only took them a few laps to catch up to the back of the pack. I also noticed that all cars basically raced at qualifying speeds for the entire race even in the heavily dense pack. It wasn't until after the pitstops did there appear to be any substantial gaps appear between the cars, only because the pitstop time varied between cars. I honesty hope that this elastic band effect is not the case, but I've noticed it in almost every race where there is never one car, or one pack able to escape from the rest. Would love Reiza to chime in and dispel these fears of mine.
Rubber banding would not be realistic in any series. Anyway, I don't think that this single "test" shows any evidence of rubber-banding. Watching a bunch of AI, AI which we all know is far from the best in sim-racing at the moment, is not really conclusive of anything. There's not enough spread across an AI field to promote much actual "fighting" and the AI doesn't seem to make many (any?) human-like mistakes, which is what keeps the pack so close together throughout. As for the ones catching up after an accident, the AI is always far faster in clean air than in a pack so it's not surprising. The AI always gets really slow in pack, especially at hairpins, chicanes and other tight spaces giving ample opportunity for others to catch up. Also, rubber-banding tends to be used by bad AI to catch the player rather than each other, to keep things "challenging" in arcade racers. So yeah, I think a lot more testing needs to be done before we start claiming rubber-banding is taking place. Personally, I don't think it is.
No, I wasn't talking about elastic banding, I was talking about "tight group races" example: realistic in stock cars, not realistic in F1. Whether that's the result of the elastic or not, I don't know
So I did the experiment again except in Championship mode. So same Stock Car 2020 at the same track. Nothing changed this time as far as the tightly packed racing was concerned. For the entire race of 55 minutes the complete field was spread across 15 seconds. Except this time no cars pitted and all cars ran out of fuel and retired with 5 minutes to go.
