*sigh*, a statement of fact about how the current system operates does not make a case for why that system is preferable - at least it neither adds to nor disproves a single thing I've said.
OK then, how about this? The current system is very simple, and therefore decipherable even with only basic mathematical knowledge (Yay! On topic!

). Which means that an average player who
- knows the various defensive bonuses (IIRC, they're listed in the manual, even if not in the Civ'pedia -- and they're certainly accessible via the Editor)
- understands how those bonuses, the A/D values, and the RNG, are applied by the game to determine battle-outcomes
- can do basic arithmetic
...should be able to do a back-of-the-envelope calculation of their odds of success, allowing them to decide whether or not to risk sending their units into battle, or leaving them exposed over the interturn. Note that the above describes a player who has taken the time to learn at least the basics of how the game-mechanics actually work (as opposed to how they
think it works, or
should work).
Sure, the
actual combat-results might not conform exactly to the player's predictions on a per-unit basis, and the short-term aggregated results might even
also fail to live up to the player's hopes/ expectations. Speaking personally, that element of uncertainty is what makes Civ more interesting than a purely deterministic game, like Go or chess. If I know for sure that I'm going to win (or lose!) before I even start, then why bother playing?
And also, the random allows for the possibility of better results than the player even dared to dream of. I mean,

must happen at least as frequently in the player's favour as vice versa (i.e. not very often, either way) but strangely enough, I don't think I've ever seen
any complaints about 'unfair combat-results' when the player's Spear is victorious...
And another advantage of a simple combat-system: when playing mods/scenarios,a player will easily be able to transfer their epic-game experience to make similar outcome-predictions for units with altered A/D-values, HP-bonuses, etc. And if they later get into modding Civ3, they can also use that knowledge to significantly decrease the

probability even further, if it really bothers them that much...