1. An even more developed sense of tactics in combat, maybe almost to the level of having a quick minigame (think playing card quick), but not to the extreme of having a nested tactical game that takes like half-hour to play out a battled (i.e. not a miniatures game).
2. More comprehensive diplomatic options. Trade more and more things, with unilateral options.
3. More resource choices. Trick is to have more while making pursuit of some resources largely optional and purely as a side strategy. So keep the core resources, but have tons more optional resources that present strategic benefits if one specializes in them. Would require some rebalancing of the happiness generated by owning multiple luxuries.
4. Automatic trade of resources, at least under certain civics. use to make certain cities a stockpile. Use worker to rush production in a city if supplied excess hammers by trade, without need to sacrifice population points.
5. More civics options. Civics are a great idea and more could be represented at all eras if a finer tooth definition for civics is applied.
6. More religions and more options with religions. Purge religions, merge religions, create schisms, etc..
7. Worker units need support to maintain, either gold or local excess happiness.
8. Add in the best facets of the Revolutions type mods.
9. Add one or two more eras, and more granularity per era (more civics, more units, more religions, etc...)
10. play around with the idea of giving great people more variety. Rather than each GS being a clone, maybe the second one could have the potential to have more reaching effects---in the same way that Civ4Col's FF's have a wide variety of effects per FF category.
11. play around with the idea of quantitative resources, supply, and quatitative logistics. Perhaps at least a partial quantitation of resources, with limits and bonuses resulting, including number of units of a specific type allowed. Require unit support to be more than just gold, but partially paid by specific resources, with a max threshold of support units being a multiplier of the number of relevant, duplicate resources that the civ owns.
12. play around with in-game modding. In game script editing, and possibly in-game model editing.
13. more & better music.
14. options for nicer graphics, an extreme view in battles
15. play around with in-city 'sims'. Not as micromanagement, but as a more detailed measuring stick of city happiness, sickness, etc... If possible, develop the idea of sickness, happiness to a little more detail. Avoid micromanagement, but make detail increases be something that master players could indirectly affect for strategic purposes.
16. play around with the idea of professions/factions a la Victoria and Colonization and GalCiv2. Again, not as micromanagement, but as something that could be indirectly manipulated for strategic game play effects.
17. Develop the economy some more. Have an option for economic wins, economic differentiation a la city/building specialization. Perhaps allow citizens to not just work tiles, but specific buildings a la Colonization?
In general, don't throw out the advances that Civ4 made over Civ3. Civ4 was a great redesign. Keep the core innovations of Civ4. Just improve them when they are too limiting or too artificial (e.g. frequent Spanish Buddhists, Monteczuma is a compulsive warmonger, etc...).
What do you think such a game need to improve from the previous one? Any ideas?