If you could boil this down, I'm not sure anyone has made a good counter. Are we essentially saying "Its Aerodromes, but less" ? Aerodromes are on one end of the continuum classical era the other. Fill in the gaps. Add in your bits about upgrade costs being higher depending on unit type and I think we're mostly there.
Okay, here goes - adding in some really Good Ideas from a bunch of other posters (you know who you are, Thank You All!!)...
Basic Unit Cap is 1 combat unit per Population Point, BUT only in cities that have a Palace and/or Encampment
In addition, the following allow you to raise that Cap:
Buildings
1.
Palace - adds 4 to the Unit Cap (This should be enough to keep you alive early in the game, in my experience: even with just one city you should be able to survive with 4 combat units plus Scouts until you get an Encampment up if needed.
2. Encampment Buildings:
Encampment District - allows you to count all the population points in the city towards the Unit Cap.
Stable or
Barracks - add 1 to the Unit Cap
Armory - adds 2 to the Unit Cap
Military Academy - adds 3 to the Unit Cap
So, a completely 'built up' Encampment with all buildings will add 6 to your Unit Cap, in addition to +1 per population point of the city in which it is located. A concentration on Military Infrastructure by building even 2 - 3 complete Encampment sets should make your empire independent of any Unit Cap - unless, of course, all your opponents have done the same thing!
Policies
Each of the following Military Policies adds 1 to the Unit Cap Per City regardless of any other military infrastructure in the city:
Conscription
Defense of the Motherland
Levee en Masse
Patriotic War
Total War
Declaration
Having War declared on you by another Civ adds 1 to the Unit Cap per City regardless of any other military infrastructure.
(OPTIONAL) IF everyone still thinks Barbarians are Too Much Of A Good Thing, then every unit of any kind, including scouts and civilians, killed or captured by a Barbarian unit adds one to your Unit Cap, but that addition is Temporary - it 'expires' after 20 turns. Continuous Barbaian attacks will 'militarize' your society, but a single incursion is only a temporary 'emergency'!
Unit Cap Penalty
For every Unit that you exceed your Unit Cap, you lose one working Specialist in your Empire. In other words, you lose the ability to work one tile or structure.
IF you exceed your Unit Cap by more than the number of cities in your Civ, each of your cities also suffers a - 1 Amenity penalty.
IF you exceed your Unit Cap by more than twice the number of cities in your Civ, each city suffers a -2 total Amenity penalty and a - 1 Production penalty.
Frankly, I suspect that if your number of cities versus units falls below 1:3 you are losing the war and the game and no further penalties need apply, but I am flexible on this: I just don't think I've ever seen an Civ at war with only 1 - 2 cities left and 9 - 10 combat units. If that situation prevailed in peacetime, then that Civ has really, really messed up its priorities!
Remember, for the purposes of Unit Cap, no civilian units or Recon units (Scouts, Rangers, Special Forces) count.
Aerodrome implementation stinks because of fundamental problems with AI. It works for me because the investment is so massive it forces me to finally choose. Am I going dom or not ? AI never seems to choose. Not even military ones. AI unfortunately plays like me @ King. Little bit of everything. Some of them should be all about Domination. They should heavily invest in dom with maybe one other thing (science/faith/culture) but instead they spread their resources around. In the early game its easy to counter Dom AI because the investment isn't large. Pump out archers. Beyond ancient the gap between an AI who invested in DOM (and lets be real, science) should be larger. So make the buildings matter more.
Does it effect ability to upgrade ? More of this
Does it effect ability to build ? More of this
Does it effect total units ? Less keen on this
Regardless of what happens there is a reckoning to be had with AI flavors. They need to be more pronounced. Having military infrastructure pay off needs to be backed up by at least a couple of AI who will invest heavily in military and then go on a rampage on the poor fools like me who ignored it because I used to be able to pump out archers/ranged and be safe. On the human side of things if we want to go domination without investment it better be done by classical otherwise the AI who did invest in military infrastructure will be ready to take their turn.
This is subjective, of course, but while I understand your argument and I agree with elements of it, I don't like the idea of playing a game covering 6000 years of history in which every civilization from the start of the game has to concentrate on one aspect of Civilization. For me, that smells much too one dimensional, even strait-jacketed. It means from the first turn I am playing only one 'type; of game, focused on only one thing for the next 500 turns. Frankly, I'd rather play Solitaire: it would be much less boring.
Also, it implies that AI flavor and Gamer Decision is independent of Game Conditions: having decided on a Domination or Science Victory, than finding several great Culture-Producing Natural Wonders to place cities adjacent means nothing? I must forego building anything that is not directed towards my chosen victory?
Absolutely not. Civilizations are much more complex than that. Even Sparta invested in Amenities, and even Athens invested in a strong military (mostly naval). Germany has been successively a Cultural, Commercial, Industrial, and Military powerhouse, and that just over the past 500 years. Over a longer period and wider area, China has been all those several times in succession or several at the same time.
If the game is going to be entirely linear (even more than it is now!) from 4000 BCE to the present/Future Era, then to me it will be a very dull game, and not worth the time it takes to play even one game.
Of course, that returns to what, I think, is your main point: to achieve that in the game, the AI 'strategic thinking' has to be improved considerably, and its ability to take advantage of the game map, conditions, and the Uniques of the AI civ it is 'playing' has to be much, much better than it is now.