Comment by chrisdirkis
Comment by chrisdirkis 2 months ago
A few quick q's, since I'm working on a game with a hand-rolled rollback impl (we have state copying, so we can do some nice tricks):
- Is there anywhere we can follow you about the clock-sync trick? I'd definitely love to be notified - On the adaptive delay, are there gameplay or rollback engine implications to variable delays? Seems somewhat "unfair" for a player to be penalised for poor network conditions, but maybe it's better than them teleporting around everywhere.
Good luck with the project! I'll hopefully have a fiddle around with it soon :)
Sorry for the slow reply!
I think maybe the Twitter might be the best place to follow for blog updates: https://x.com/MadeWithEasel - I will definitely be posting about every blog there once I get there.
On the adaptive delays: I have a multiplayer game which gets about 100 players a day and it has been interesting seeing how they have all reacted to various iterations of the netcode. The overarching thing I've learned is that latency is quite psychological, it's the difference between expectation and reality that matters. In other words, high latency doesn't necessary mean an unhappy player, if they are expecting the high latency.
First though, the amount of rollback is limited to what the player's device is able to handle (there's yet another algorithm I've made for collecting statistics and estimating this!) Some devices cannot handle any rollback at all and so unfortunately sometimes rollback netcode isn't solving anything for those players. I think these are the cases where the adaptive latency is more important.
We sometimes would have games where we have 3 people from the US playing happily together, and then 15 minutes later 1 person from South Korea would join, and the latency would jump up dramatically for everyone. The US players would feel the difference and become unhappy. The simplest way to explain what Easel does now is it places the server at the weighted-average midpoint between all the players. So in this case, you can imagine that the server started in the US, and the moved 1/4 of the way towards South Korea (since 1 out of 4 players are in South Korea). I have found this to be the most fair and the key thing is the players find it to be fair. It matches how they think the latency should be apportioned and so they are okay with it.
Recently though I added a feature which splits the world up into regions (it's more complicated than it sounds because the regions flex around the players a bit, see https://easel.games/docs/learn/multiplayer/regions). By default, you only play with players near your region, but you can switch into Roaming Mode and play with people all over the world. The trick here is, when the player chooses Roaming Mode, they are explicitly choosing high latency, which changes their expectations. When they get high latency, they expect it, and so they are happy. The funny thing is, the algorithm used to automatically assign them the same high latency in these situations but players didn't like it because they didn't have the choice.