Speaking strictly IMO:
1) Looking at LoV. I would have loved they simply ported a version of FFT's engine into the game and let us use our minions on that. Let us 'customize' their abilities and 'level' them up against AI and play with/against friends. To supplement this idea, they should have a tile editor for people to create and share maps and scenarios. I literally would play this nonstop, especially if they tied in other rewards. Alternatively this could have been a squadron feature as well instead.
They even could have just gone Pokemon battles and it would have been 'sufficient', but I think drawing on inspiration from past successes and IP is a better way to go.
2) UCoB is great. Except one for glaring issue. 18 minute fights aren't fun. They weren't fun when I was progressing world first shit in WoW and they're even less fun now that I am older. They should design boss fights to be shorter and more impactful, not longer and more tiring.
3) The Glamour Dresser is a perfect example of one step forward two steps back. They should have just not touched it or spent any dev time on it until something actually good could be implemented.
Now to supplement my post with some of my own ideas:
1) Gold Saucer is downright criminally underused. Chocobo racing in its current implementation should be a felony. Much like my LoV suggestion they should have ported in a Mario Cart (or better yet Crash Team Racing) engine and let you race AI or friends for prizes. I'd have tied the entire Chocobo progression system, bardings, colors, and EXP/leveling/effects in here as well as other rewards (tomes, gil, materia, etc.) Then supply the playerbase with a tile editor and allow us to make our own maps to create and share with friends.
2) Battle Arena could have been a really cool idea if we had a slightly more robust battle/gearing system. Something like a "create a boss" thing that gets assigned a random model, random mechanics, etc. and they ramp up the longer you go on with a wheel to spin to get random debuffs/buffs to increase payout.
3) Regarding Eureka. I made a post in the exploratory missions section of this forum. It was basically a procedurally generate map using a random tileset from all available dungeons/zones and dumped you in random spot on the unexplored map. Thematically, it was the idea of the 'item world' from Disgaea where we delve into the Relic itself to empower it.
It would have random monsters in the map that had "affixes" a la Mythic+ from WoW. These monsters could have mechanics that need to be interrupted (a la CC), kited away from other mobs, killed in specific order, flee and attempt to grab other mobs, random threat drops, or advanced AI like harassing healers, or focusing down players who suddenly get low on HP, etc.
The content scaled from 1 star to 5 stars, with 1 star being similar in difficulty to current dungeons. In 1 and 2 star difficulties mobs would have their 'affixes' listed above their nameplates and give context into their difficulty and if they were worth fighting or sneaking around. At 3 stars+, affixes were hidden and were only discernible when fighting the enemies.
You would clear enemies to fill a bar that spawns a boss in a random location on the map. You could then go hunt the boss, or keep killing mobs to spawn yet another. Bosses would have a random model (think like my Battle Arena concept) and randomly assigned mechanics. Availablity and quantity of mechanics would be directly related to the star difficult. I.e. a 1 star boss might have a tank cleave/buster, and 2 isolated mechanics. It'd be equivalent to fighting a dungeon boss at say min ilvl, with maybe marginally higher outgoing damage. At higher stars bosses would get more dangerous mechanics as well as multiples and they may begin overlapping. I.e. a tank cleave that needs to be shared, and lowers max HP on hit.
The idea behind this would be repeatable battle content that scales for all player skill levels while offering meaningful engagement and rewards. In addition to mobs and bosses there'd be treasure maps to explore (they grant an instant boss spawn as well).
In summary - the idea is that there should be better battle content that scales for multiple player types as well as better side content that is well designed enough to justify its inclusion and be fun enough to keep people playing it as well as offer viable alternatives to rewards.
To be fair, I find the hivemind to be unsettlingly defensive of the game and completely against any and all criticism. It's exceptionally common to see 'go play x instead' instead of actually debating the quality of the persons post.
I agree on Coil stuff thats why I snipped it. But I wanted to point out that I agree with your analysis there.
On the other bits though, I REALLY enjoyed Leviathan EX. It had the right kind of randomness and RNG and made the fight really fun. I also liked Exdeath Black Holes, but otherwise agree his mechanics were boring. I actually really like Trine. I think it's a good mechanic, but otherwise agree that pretty much Primals Shiva and further were way too scripted or generic. I really loved the control aspect of Moogle Mog, Levi's randomness, Ramuh was greatly designed, and the original 3 were all fine and dandy because they were different. I also REALLY enjoyed second coil. I felt like the content was a little less scripted and more dynamic.