AE and coalitions are fun in the way that death is fun in an RPG or FPS, and that crashing is fun in a racing game.
They're not fun. It's the work required to mitigate and avoid them that is fun.
No. Those analogies do not work. You do not die in RPGs from doing too much damage. You do not crash in a racing game because you're half a lap in front of the opposition.
Coalitions are much closer to rubber-banding (IE AI cars getting a speed boost when behind, but losing it when they get close again) and have no true correlating element in RPGs whatsoever, unless you want to count NG+. Unfortunately, rather than rubber banding when the player is in the lead, they currently rubber-band when the player advances from last place to middle of the pack...and rubber band less the further you get out front.
You'd be looking for dormant cores and low-AE means to expand constantly, and take risks when it's worth taking: do you vassalize the Hansa in their 1444 border, just because they joined the wrong side of the war and take the AE? ...or do you let the Hansa grow, take extremely rich coastal cities and try to dismantle it province by province with infuriatingly high war scores and AE later?
Problem is, with the gridlock given right now, you don't have a choice. You just avoid coalitions all day every day until you're ready to chain them for breaking truces w/o penalty. People list *the* strategies for avoiding coalitions, and they're static: start with wars that minimize AE, like tearing out Styria, Burgundy (if inherited), French minors, Sweden/Norway after integration. Then, you're too powerful to stop and can just chain everyone else alive to death. That is, as Jomini says, "strategically bankrupt". Virtually every nation in Europe will do the "return core feed + colonize or let vassals colonize" in SP. Even in the Muslim world, that's how it goes unless you're a horde, or the Ottomans where you basically start able to hammer through coalitions right around the time where you get the full-annex mamluks mission.
I'm OK if the coalitions stay. Sure, you can argue that the mechanics could be more fluid, more interesting to work against rather than around. And those points are probably valid in many ways.
Creating the conditions for coalitions or not is a choice. Currently, it is a false choice unless you want to use them to break truces for free. Dynamic gameplay elements offer actual choices, not shoehorn linear progression. Coalitions do the latter.
IMHO it's better to stick to one less-than-ideal-but-working design than to go bipolar.
They are not a working design. Most people claim them to be an anti-blobbing mechanic, but that only applies (and only sort of) to small nations. They are a punishment mechanic, and a poor one with easy workarounds. They would function a lot more as the old school "dogpile the runaway player/nation" mechanic, and it's notable that that is EXACTLY how they used to function with size scaling, though it led to some wonky stuff. Their current format of only punishing small nations in any material way is an abomination.
The coalition mechanic exists to slow down player rate of expansion, and to provide more challenge for successful players. Without coalitions, and even with them for easy-mode nations like the Ottomans, it's currently possible to faceroll your way to world domination with most nations by 1700. Coalitions provide a threat to players (and sometimes to the AI) who expand too aggressively, and forces them to consider the impact of their actions beyond whichever country they currently happen to be dismembering. Even then, it's perfectly possible to avoid coalitions provided that you don't expect to be able to directly annex vast territories in a single war.
Well, the part about being able to faceroll is accurate, as is the last sentence, but the rest of this paragraph is not.