1. Have intermediate steps between "uncored - this breaks the empire" and "cored - it our patrimony forever". I like the idea of splitting land you own up into: unclaimed, unrecognized claims, recognized claims, and cores. Unclaimed land acts as a short term brake on expansion - you get OExt, it revolts easily, gives less money, and it has lower defensiveness. An unrecognized claim means that you have some shady claim on the land, but the populace doesn't really think it is valid - you lose the OExt and some of the revolt risk, but this land isn't something that really powers up an empire - it is still being digested: this is a long term brake on growth (large states typically take more provinces without claims than small ones). A recognized claim means you have administratively integrated the place - you now full money and lose some more of the RR and get more manpower - this represents that the people don't consider themselves full citizens of the state - but do recognize their place in the empire. A core would have full benefits for everything - finally getting back to normal defensiveness and dropping the RR. I like the progression: unclaimed (you took this in war without a claim) -> unrecognized claim (you took this with a claim on it, it defected, you bought it or it was unclaimed and you had diplomat generate a claim - with time depending on wealth/culture/religion) -> recognized claim (you diplo-annexed it or you had a claim and then spent ADM to core it) -> core (you integrated a core vassal or PU or you had a recognized claim and then waited a long time). By breaking out the functions: this causes OExt and prevents mass short term expanion, this prevents this province from jacking your income/manpower, and this gives you a fully integrated province we can a lot better balance the early game vs the late game. Some government types/ideas/decisions might make it easier to hold say unclaimed land (yeah early expansion while I'm a crappy despotic monarchy), while others might make it easier to hold say unrecognized claims (yeah I'm a revolutionary republic - just let me indoctrinate the masses). This way you can keep the game dynamic - up until a certain size you can easily and quickly push a province all the way to core, but as you get larger it becomes less and less cost effective. So now new conquests contribute less (wahoo less snowball effect), but the core empire doesn't revolt if you say take French North America. Province status - unclaimed, unrecognized claim, recognized claim, and core - could be used as an event trigger or for mechanistic purposes. Right now we run into the problem that we have a pretty strict binary so something like the Ottoman conquest of Egypt gets treated exactly the same as a Russian conquest of England. Allowing intermediate states can let us set up things so some of the major conquests can be better represented (e.g. allow colonial powers to start off with unrecognized claims on colonial conquests). They also allow us to say tweak downward the value of new provinces (hey these lands with unrecognized claims are worthless), without making a wall that breaks the country (crap OExt just hit 200, I'm doomed).
2. Get rid of the percent warscore system. Right now you have two basic approaches to war: win battles, take goals, and blockade or siege everything. As siege everything gives 100% we end up needing the war score system to twist itself in a knot so that something worth "5 major battles" scales well with say 1/5th of total occupation. Instead I propose having province occupation, battles, and all have fixed values. So province A and province B have the same 20 and 10 warscore regardless of who owns them. Now the balance part - when making peace concessions follow geometric scaling. Say A costs 20 and B 10, right now you pay 30 total for both. Given the flat overhead for war - it makes to keep beating down the loser to take A and B as the cost for wracking up war score to B is marginally cheaper than taking A was. With a geometric progression, each additional concession takes say 2% more than the previous. So now for A & B you need 32. Or for three 20 cost provinces you'd pay 73. This means that big wars late game (win or lose) can have big changes, but not if you just game the system for a quick win. Against an anti-Napoleon coalition you might snag ten 20 cost provinces - but only because you've wracked up 520 war score in battles and occupations.
3. Expand the vassal system. Right now you are either independent, a vassal, or a PU. This misses a lot of history and makes it hard to balance relations. Instead I think we should have: tributary states, hostile vassals, friendly vassals, integral vassals/PUs. Tributary states pay you gold monthly (so the AI doesn't get themselves creamed whenever they are tight on funds), if they stop paying (click a box) - then the overlord gets a CB to go take some land or vassalize. Hostile vassals are vassals you got via force. They actively look for ways to backstab you. You can trust them - if you are stronger and aren't busy elsewhere; they can make deals with foreign powers to betray you in exchange for foreign support towards independence (if they try this and fail, you can convert them to friendly vassals). Friendly vassals are vassals you've DVed or have spent a long time furthering their interest. Integral vassals regard themselves as nearly autonomous regions of the mother country - when you annex them you get free cores in the previously described option above (basically they are junior PUs without the monarch sharing mechanism); they also do not require a diplomatic slot. This means that you can't just dump land to a vassal and forget about it. As vassals can migrate between integral/friendly/hostile you can't just annex the whole crop and then forget about it. Having integral vassals is a good way to get cores (though at the expense of your vassal relations elsewhere) or a good way to control land without owning it (though now you are constrained by keeping the vassal from slipping down the ladder).
Right now, vassals are pretty much non-entities. You keep them until you need to integrate them (unless you abuse the HRE). Vassal balance runs into trouble because it has to cover a lot of historical ground and making hard on the big boys to manage or annex multiple vassals almost always makes it fiendishly hard for the midrange powers to utilize vassals well.
The main problems I see in the late game are:
1. Choices die as mechanics become straight jacketing, there are only a few ways to play and you have to adopt playstyle to avoiding mechanisms.
2. Anti-blob mechanisms make the game unfun for weaker players, cranking up the anti-blob makes it really hard to achieve lift-off for early games.
3. The end game is just a repetitive giant status quo preserving racket. You war, you take 4 provinces, you wait - lather rinse repeat. Using feedback loops (like rebels) means that you have a very narrow range where you can play so the end game plods along boringly.
I think a lot of these things happen because mechanisms double dip. Breaking it out gives you more choices, and lets us fine tune where and win to make pushback mechanisms.
