Capitulating Masters of Vassals - Bug or Intentionally Horrible Feature?

[TheMeInTeam]

Let me paint a picture here.

You have more than 20 cities and well over four times the power ratio of the civ you are at war with. You yourself have a vassal that's just under twice his power as well. He's down to two cities because you took six and your vassal took 2. So, he's got 1/5 of his cities left that he originally had.

He won't capitulate.

Whoa. Now hold up, why might this be? Well, because he has a vassal. A vassal with three cities way out in the middle of nowhere that also happens to be weak and backwards. In fact, your OWN power is triple their combined power, putting aside the aforementioned vassal you have that is easily stronger than either of them and just barely weaker than both combined on the power charts, which should make it even stronger.

No no, still won't capitulate...just a tech.

I've noticed this being a consistent problem - civs with vassals, regardless of whether they're colonies, voluntary, or capitulated vassals, are much, much harder to capitulate than civs normally. The problem here is that if you took the civ + vassal, turned them into 1 civ, and did the exact same amount of damage, it'd have capitulated long ago.

My question is this: is there something unseen in the code that makes this occur, or is this actually coded as intended? If the latter, I'd have to combine a question as to what the person writing it was thinking, along with several expletives, because this is garbage nonsense.

I've heard rumors about the "that goes against everything we stand for!" thing being intentional too. Seriously? The AI won't swap out of bureaucracy because it likes organized religion, and that's INTENDED? I better not digress too much though. What do people know about this vassal state inflate power pretend nonsense?

 

[beestar]

I suspect the AI calculation weighs costs and benefits, and the costs of surrendering to you are very large. Not only does it become your slave, it also loses its existing vassal (that vassal either becomes free or joins you; either way, it loses that land/control).

One experiment to prove the argument might be to use different sized civs as the existing vassal. The objective would be to prove that it's very difficult to convince the enemy civ to surrender when it owns a useful existing vassal, and that it's easier if the existing vassal is small/useless/resourceless.

 

[Dresden]

The estimation of their power used in deciding to end a war does include vassals; it is intentional so far as I can guess since it is an explicit choice. As to how big of a factor that is though, I can't tell you since power is one of many factors.

It would be an interesting test when you face such a situation to go into worldbuilder and take away their vassal (through the diplomacy window) and see whether they'd choose to capitulate without it and, if so, how much earlier.

 

[DanF5771]

The problem with AIs "doing fine on their own" is that there are actually 2 tests measuring power ratios when it comes to capitulation:
1) compare power of the AI vs. your power
2) compare power of the AI vs. the average power of all players alive in the game

Both tests must be passed, otherwise the AI will always respond with the above mentioned infamous DENIAL_POWER_US. So I don't think it's a bug, but something to be aware of in order to adjust one's strategy to in a particular game. And I'd rather get a different denial like "We are among the strongest nations in the world" instead.

An AI with a vassal will be above the average power level in most games due to the "power" that comes with technologies and national wonders being counted twice. Maps with a lot of small islands and many 2/3-city civs spawned as colonies need special attention / an adjusted strategy. It's better to kill civs completely than to keep them in the game as your vassal with one garbage city somewhere near the north pole, thus dragging the average power level down.

It's also good to know that the value for the average power will be "adjusted" by additional threats to the AI not willing to capitulate, if it is a land target for other stronger players (even better if those players are your allies in your war vs. the AI, but be extra careful then -> "You have made peace...", "AI has agreed to become vassal of XY..." ). Better conquer islands first / spare them the particular city on the particular landmass which lets them remain a valid land target for player XY.

 

[TheMeInTeam]

It's late and I'm going to bed, but these don't sound like plausible explanations to me. Fortunately, I held onto a save for this one, so I'll post it from my other computer tomorrow.

Again though, we're talking about a 2 city civ (that lost 8 cities to get there) and its 3 city vassal, against a 30 city empire with four times the power of the target. FOUR TIMES! I don't know what tiny fraction of my land and pop they had combined, less than 10% most likely.

Interestingly, DanF may have hit on something with the averaging. I was far and away ahead of everyone at that point, the target AI would have probably been like 4th or 5th out of 12+ civs if you combined its power with its vassal.

There's nothing in the code concerning vassals when it comes to this Dan? Even later in this game I noted a spectacular difference in civs who didn't have one. In fact, at the end of the war (which once I post the save you can simulate also), TWO OTHER CIVS OFFERED VASSAL STATUS PEACEFULLY. One of them was *notably* STRONGER than the guy I'd just been fighting. I mean the beating in the non-capitulation war was just flagrant! Also Kublai Khan on this map vassaled after losing one city, and he was far stronger at that point than Pericles had been...

So maybe my question's been answered already (Intentionally HORRIBLE feature it is, then?). Something just feels awkward about having three or more times a civ's power, ripping it to shreds for over half of its cities, and "it's doing fine on its own".

 

[DanF5771]

I got my information from CvTeamAI::AI_surrenderTrade. The powers are combined for Master+Vassal for the target AI who you want to capitulate, but not for you and your vassals (well that didn't matter in your case as you were powerful enough alone).

But 12+ civs in the game and the target at 4th or 5th rank by combined powers is exactly the situation where test 2) fails ...
You can't do much about it now (your insane power advantage doesn't matter much for test 2) other than kill other weaklings quickly to raise the average power level. Any chance for the "land target" factor?

I don't like it very much either, but it makes you think twice about capitulating a player vs. making him leave the game completely.