1. May seem trivial, but a random map size option - preferably executed in such a way that an 'Advanced Options' version allows you to choose which map sizes are allowed (so, for instance, I could disable duel maps while allowing any other size).
2. Individualistic AI personalities a la Civ V, and the attendant removal of the agenda system. Even if the first can't be provided, the agenda system needs to go or at a bare minimum the reasons for denunciation etc. need to go back to being hidden. The gameplay combination of civs with strongly stereotyped attitudes along a single axis and indistinguishable behaviour from one another is cripplingly counterimmersive, and being told I'm being denounced for moronic reasons only makes matters worse. Even when it makes conceivable gameplay sense that a civ would choose its rivals based on perceived weakness due to a lack of walls, small population, low productivity or whatever, it's utterly immersion-breaking to be told that's why they're doing it, never mind the agendas which have no bearing on gameplay position (wrong sex, not settling foreign continents etc.) or actively favour hating stronger civs.
3. Remove the religious victory and combat systems. Religious victory is tedious for the player both in terms of its core mechanics and the need to overcome AI religion-spam behaviour; it presents no challenge in terms of a risk the AI will achieve a religious victory as in any game with more than one AI religion this isn't actually possible; it's utterly counterimmersive (at the very least there should be either a global or an era-based cap on apostles, and once they're gone everyone has to use missionaries - most religions have very few apostles. Gurus have only stretched credibility further); it's only available at all if you found a religion (no other victory is tied to something you need to do in the first 60 turns); and it's completely unrelated to any other game system.
As an extra one, victory conditions need reworking across the board. The cultural victory conditions are opaque given the lack of interface clarity on the relation between culture and 'domestic tourism', it's ultimately a score victory as the tourism resource serves no game function, and simply in design it's just a "Fill Bucket X" victory just like science or, in effect, religion, only shallower than either since filling the bucket is literally all you need to do - no converting faith into units that convert cities, and no attendant project to complete that demands a separate resource, as science victory does production.
Domination needs to go back to the 66% global population threshold rather than the 'control all capitals' system as the AI is literally incapable of winning a game that requires it to take out every AI capital plus the player's, since unless the player is the last civ standing most of the time the player will be out of the game before a domination victory can succeed.
There's nothing intrinsically wrong with the science victory, though the tools to make it easier added in R&F (builder charges or gold-buying) should be removed. It's the most well-rounded condition: science being linked to pop intrinsically rewards playing the game in a way that manages expansion and city growth well, it demands good production as well as good science, and it's strongly tied to the spy and - previously - the Great Person system (it's now too easy to win without any of the relevant Great People). Still, it's a bit too easy to simply 'stumble into' without advance planning of any kind, since the structure of the tech tree is such that you're going to get most of the relevant techs naturally where in older games you needed to specialise further in advance.