I'd recommend applying to any Civ V changes some of the lessons-learned from the half century of recreational board and miniatures wargaming, and also some from the real military simulations world. A couple of the most important are that we cannot directly transpose game mechanics from the board or miniatures genre directly into computer games, and perhaps more importantly, that the game design and mechanics needs to avoid mixing strategic, operational, and tactical concepts in one game format. Civ V, for example, is excellent for representing tactical conflict at the tribal, city-state, and up to the duchy level, but the mechanics fail at the operational and strategic engagements of the nation level, so it shouldn't even attempt that game scale. Thus the recommendation to trim the tech tree, units, and such to end at the rise of large national armies, as in the Napoleonic Wars. At the game distance scale, with single tactical unit per hex, a single nuke would wipe clean the entire map - a nuke, heavy bomber, or ICBM is out of place in a tactical one unit-per-hex game. Another point tactically is the use of melee units to protect ranged-fire units, for example including some spearmen or pikemen with archers, or some musket infantry with cannon. So it would be nice to be able to stack a single melee unit with a single ranged-fire unit. It would also be nice to replicate some historical tactical elements, such as forming an infantry square in the face of impending cavalry attack; also the availability and correct tactical use of light cavalry/hussars, mobile ranged fire weapons like horse artillery, and light infantry advantages in rough terrain. And of course, a smarter AI, especially one capable of selecting and implementing a particular battle doctrine would help allot more than biasing combat resolution against the human player (or am I thinking of Civ3).