OK.
Basing the entire economy of cities on special resource tiles makes these tiles less than special.
When I read that, once again, yet another resource would be seeded next to every starting position (a shard this time), I started to wonder... as time goes on and the game continues to be uprooted and re-planted, will there will be any room left for a city in the starter clump of these "special" or "rare" tiles?
One of the interesting things in the game is how the world is changed based on being in your area of influence. This suggests that the terrain is improving, becoming more fertile and productive, or whatever. Unfortunately, this has absolutely no affect on anything. It's just some pretty visual effect.
If you're going to add cool spells that let me wipe out huge areas of land using your touted deformable terrain engine, you need to make those changes worth a little more. If I destroy or screw up half of the land in your city's area of influence, I expect your production or economy to suffer. It got all pretty and green when you moved in, now it's all on fire. Both sides of this coin need a gameplay ramification.
This brings me to the title. The idea that everything is based on "special" tiles is, well, a bad one. I can only assume you did it to discourage city creep, but, well, it doesn't help. It's still most effective to spam cities. As long as there's shared pools of resources, each new city is another unit factory, and more of those means I get a bigger army faster, whether or not each city makes more than a gold per turn.
You keep throwing bandaids on the train wreck victim that is Elemental. A broken jaw, three compound fractures, and massive internal bleeding isn't going to go away with some gauze.
I understand you removed worker units at some point, which was nice to remove micromanagement tedium, but bad in that land essentially doesn't have inherent value. We can't improve empty squares, they're just... there. Only the special tiles matter, and that makes them necessities, not special rarities.
This is an expansion B suggestion.