y'all are making animl look like the reasonable one.
Let's simplify it using that example video.
Collisions are determined locally. Each client determines their own collisions.
If you are suggesting rewriting Aces High to use an Authoritative Server design, well that is a complete rewrite with all the emergent bugs that might introduce. Good luck convincing HT to go that way.
Now let's take it from that jug's point of view.
If you try and project a ghost plane across the last known vector to determine collisions, speeds can change, internet lag can fluctuate, the player could have changed directions since the last packet you received.
Some projection was already being applied. HT had a patent on that. He understands how that works.
So the jug was already applying position smoothing from the last packet he saw. But that still put the other plane ~6 plane lengths behind.
You want to apply a phantom collision only to get the next packet and see the other plane had peeled away? Or reduced throttle since the last update you received?
The further out you try and predict\project, the larger your error becomes. You predicted a collision out too far based on a straight line assumption and found out in the next packet he did not maintain that straight line in that gap.
It would be hard to undo a collision you predicted and applied before you got the update the other plane swerved away.
And why do you think you have to convince us? This is not something we are going to vote on.
Now if you want, you can admit that in a peer-calculated collision model, this is the best trade-off you can get without introducing even worse problems.
Instead you can reframe the suggestion to drop the current design and redesign AH as an authoritative server. That narrows the discussion and doesn't suggest there is some magic that can be applied to get a better peer-calculated collision model than already exists. That is a totally different discussion which can be had.
