I've just read this full thread -- All I can say is: Expect some seriously spurious "history".
It's already there -- easy, current example: "Medieval Infantry" is utter nonsense as portrayed. At best it portrays English knights fighting dismounted at Crecy et. al., and brings to mind some "Prince Valiant" type images of castle walls being scaled etc. ad nauseum.
Of course, in the developers' defense, one can always argue that there must be a balance between historical veracity and complexity / cost / appeal / etc. That having been said ...
1. "Enkidu Warriors" is utter nonsense; Enkidu was simply the hero's pal in the Epic Of Gilgamesh.
2. Cataphracts of any sort would of necessity introduce the concept of horse archers -- which (i) would be marvelous for steppe peoples and all that, not to mention that (ii) the cataphract was an evolutionary response to the threat posed by these same steppe horse-archers and (iii) would probably tie the AI in knots (yeah, I'm just being cynical here -- but the fact that such units don't exist, given their extreme historical importance, is telling).
That having been said, it'll be interesting to see how Civ addresses the cataphract issue, as some sort of Byzantine heavy cavalry is certainly required.
3. Personally, I'm expecting a major debacle on their attempts at pre-Columbian Civs. For an inkling, see
Images Of Aztec And Inca Warfare. (i) Note llamas being used as pack animals to increase mobility; will llamas be a strategic resource as such, or a luxury resource along the lines of wool? Stay tuned ... (ii) those Aztecs wore some wild "uniforms" ... don't expect to see them replicated (iii) a very important Aztec weapon was the humble spear-thrower; again not holding my breath etc. to see how these units are(n't) portrayed.
--On the bright side, I think I read something about terrace farming in the pre-Columbian scenarios.
On the REALLY bright side, I also read that every new feature implemented for the scenarios will be echoed in the editor.
As for new terrain tile types, these should be simple software objects, without any need to replace one tile type with another.
All for now --
Oz