> Action implies something changes (there is a state called "before" and state called "after"). But nothing changes "over there", we just now know what actually happened (there is no state called "before" or it's value is "unknown", only state called "after").

> using "action" suggests that something changes

By my understanding, the delayed part is important. It implies something did change as the state of the earlier entanglement will be different based on a measurement which occurs after the entangled state has been created. If this is correct, then there has been an "action" at a distance.

Though an alternative understanding could be that the state is set based on a future event which hasn't occurred yet. That would avoid needing an "action" to change the state after the entanglement event, but implies other things.

I suggest using "spooky correlation" to better explain this. It's just a forced correlation of two parameters after all. To make entanglement, we make an experiment where we force system to behave in a way that correlates results of two parameters. When we measure one, we know the other is correlated in a way that was set in our experiment. "Spooky" because it can hold at vast distances and across time.