But the effect is very much real. Much more real than you might think.
You can't use it to communicate faster than light, but you absolutely can use it to coordinate faster than light. It's called quantum pseudo-telepathy. See for example the Mermin–Peres magic square game. Two players with two pairs of entangled particles can win this game with 100% probability even if they're lightyears apart, whereas players without this resource can't win 100% of the time if they're separated (e.g. by adding sufficient distance and a time limit to the game).
If you were to use this explanation to anyone that isn't already deeply familiar with how quantum entanglement works, the "instantaneous" part of it would absolutely imply either FTL communication or at the very least the idea of a hidden variable behind the particles at the point of entanglement. And that's assuming they even understand what an "eigenstate" even is.
It absolutely is a FTL effect, even if we can’t use it to communicate classical information.
I disagree.
Ask 10 random people to read the phrase "links them so that changes in one affect the other", without any other disclaimers attached.
I bet some of them will think that means you can change A and see the change on B.
It's not good wording.
2. There's no tradeoff here. "more important" doesn't matter. It's easy to make a sentence that states an FTL effect occurs and that data can't be sent.
It also assumes a specific way of representing the quantum state (eg a state vector over all parts) whereas other representations like Feynman paths have no corresponding notion of an instant change.
What we can objectively say is that entanglement doesn't care about the distinction between time-like separation and space-like separation. If two non-communicating parties are entangled and they perform an experiment, there is no observable distinction between the cases where one of them goes first or the other goes first or in whether there is time for light to travel between them in the delay between their two parts. But that is much looser than requiring the effect to be instantaneous in some unspecified frame, because only requiring that there's no observable distinction is a much looser constraint than specifying that it must be instantaneous. And the only thing that we can experimentally verify is the looser constraint.