I hate RNG with a passion. AI RNG is fine as players are unpredictable too. But RNG in combat is ****.
Combat should not be predictable. If a superior force is caught flatfooted, they can still be defeated. That is represented with randomness in this game - however inconvenient it may be in some situations. I can imagine better systems to fight battles, but without a random factor, you could just give up as soon as you see that a strength-15 soldier approaches your strength-12 soldier. And you can still beat the numbers with more numbers. Say, ten archers attack your city defended by a machine gun - the odds are heavily in favor of the lone gun, but you should have placed a second gun there just in case that some of those archers get lucky shots.
1. The current military civics need some polishing, IMO. The only civics available until colonialism (that's in the renaissance age, mind you) are conquest and tributaries, which both have big economic penalties. Often I would not choose a military civic at all till I get access to colonialism or standing army because the cottage growth -50% and no pop growth while producing mil units are such big penalties.
I don't see the problem - you need to be creative with combinations, imo. When you're a coastal civ: Combine republic and conquest. Yes it means that cottages and farms are worth less, but you can work the coast tiles instead of the land ones, and buy more experienced troops. Use them to conquer foreign cities, switch to Despotism+Caste, improve all the land and whip barracks and other cheap vital buildings there while the town grows. Once the captured foreign cities threaten to outgrow your core, switch to tributaries and monarchy: Now produce all your troops in foreign cities and don't worry about the cities growing - the new troops will keep you happy or can actually be used. The trade route bonus doesn't hurt.
So yes, in the early game, I don't work much with cottages - they are going to suffer in the next plague, anyway. I rather start all the cottaging before Individualism comes around.
And Despotism is not as useful as it looks. Whipping converts 4 population into 120 hammers and 2 unhappiness - vanilla only assigns 1 unhappiness in a slavery-whip. So, Despotism is best used for smaller projects that allow 2-pop-whips. In RfC, I whipped factories and airports!
It all depends on the situation, of course. Isolationism can be very worthwhile when you wouldn't have any good foreign contacts anyway, say with Japan or the Tamils when surrounded by Independents. And Cottaging IS the way to go when you're playing Mali, so Conquest is a no-go for them. While Conquest is really great for Rome and Greece. And Colonialism is of limited value for non-colonizers like Mongols or Prussians who instead profit greatly from civics like Tributaries and Nationhood. Clergy is THE civic for manic infrastructure builders like myself, but when I play Moors, I need Monasticism to get the GPP. And conquest-reliant civs as the Arabs or Spanish will profit from Theocracy...
IMO, the pro-and-con weighing civic system of Civ4 (including RFC, DoC, RI) is vastly superior to the bonus-stacking system of Civ5. Actually, the reason why I came back.
[Edit/correction: I now conclude that the whipping formula is [whipped pop]/2 = [result unhappiness], when posting I assumed [whipped pop]-1 = [result unh.]