| > Fizzi released a dolphin build with rollback netcode ... making it a game with better netcode than most current big name fighters. I see this repeated often without justification. Why is rollback better? Based on my basic understanding, rollback netcode would absolutely fall apart under any suboptimal network conditions. If there's too much lag or packet loss, you and the opponent character will start teleporting all over the screen, missing lots of inputs, etc. It makes sense that rollback netcode would be preferable when paired with a very good, reliable network (e.g. so if any rollback is needed, it's only a frame or two). But if you're selling a casual game meant for anyone to be able to play, a delay based netcode makes more sense to me. An average Joe can understand "the game is buffering", but if the game starts rolling back too much and you can't control your character because he's teleporting all around, that just looks like a broken game. An interesting follow up question: if rollback is objectively the better approach, why is it implemented less often? Is it significantly more difficult to develop and does that offset how much value it would add to the game? |
The reason rollback is rarely implemented is because it is hard to motivate a company to do it when the benefits are so intangible. “Better netcode” doesn’t really sell games, flashy graphics does. And since most of the FGC has existed offline, online is just an afterthought.
This video explains fairly well: https://youtu.be/1JHetORRpfQ