I thought it’s extrapolating, because it’s not between two given frames (but on the other hand it’s definitely between known states). Let’s just call it prediction.
The Server cannot predict the state, which is the problem with all of this.
It’s about a robotic gripping arm (is there a special word for this, inserter?). This arm moves items from A to B, like this: First it moves from A to B, at B it waits a given time and then puts the items into the target inventory, moves to A again, waits and removes items from A’s inventory.
The catch is the Power Grid: When there is more power consumed than produced, there is no fuse flying, but instead every building/arm moves relatively slower. This means, that the time it takes from A to B depends on the state of the Power Grid at every simulation timestep.
One could argument that I can build a logic into the system, so that it sends said grid load factor everytime it changes, but that does only help at a small scale, because for a big factory, you have many arms starting/stopping moving at any random time, and thus the load factor is ever changing.
This however means that my prediction fails when the power would be going down, but that’d be corrected every 200ms and isn’t a desired play state anyway. I don’t know if I can get away with that, but I guess so. I could build a system which only updates the load factor when it’s over 1, so packet flood only occurs when the energy system is overload, but I’d rather solve that on a general purpose level, because similar problems might occur for other things (especially conveyor belts, where individual items move and can get blocked etc)
Oh and: For the Wait time, only one packet is sent, so it’s definitely only the movement, where packets are flooded.