AI Capitulation Calculus? (K-Mod)

[AspiringScholar]

It's been my consistent observation that, in wars between the AI midgame onwards, the losing civ tends to capitulate swiftly, so that the leading AI tends to collect numerous vassals in pretty short order, however, when it comes to a war between an AI civ and myself, annoyingly I end up having to conquer them almost completely before capitulation becomes selectable in the diplomacy screen, at which point it makes more sense to just go ahead and completely conquer them anyway. If I begin a war with the intent to grab a vassal, it would be helpful to know what sorts of factors ought to be anticipated or manipulated ahead of time in order to make this feasible.

Does anyone have any specific info on the AI logic with respect to capitulation? A quick disclaimer is that I play Realism Invictus, which has integrated the Karadoc mod, but otherwise has made virtually no changes to AI diplomacy. If I'm not mistaken, K-Mod is one of the "Orthodox" ones within the hardcore community, and hence I've come here to consult the seasoned experts, if anyone can help me out.

 

[drewisfat]

No, K-Mod most certainly is heretical. One day when our rightful emperor reclaims the Throne of Justinian, reunites the five patriarchates and Orthodoxy reigns supreme, we might consider granting re-baptisms en masse to the wayward K-Modders. Unfortunately at this time we must consider them apostate.

Sorry errr... K-Mod makes no direct change to capitulation, RI might. What's going on is a result of other KMod changes, namely that the AI builds a lot more troops and keeps them alive (regrouping as opposed to suiciding). More units = more military power = the AI doesn't want to capitulate. Other AIs will also have very high unit counts so it's not an issue between them. The culture changes in KMod (which favor the AI in practice) will also make it more likely that they land target each other (8 border tiles) than with you relatively speaking. The hesitation in committing to a battle it's unlikely to wins causes the AI to regroup with more forces before attacking. This means sometimes AI-AI wars are complete stalemates (not even fighting), but compared to BTS a 60/40 power dynamic is more likely going to result in the stronger AI steamrolling. The other culprit is the change to AI dogpiling behavior - AIs are much more likely to DoW someone already at war, sometimes declaring war on a dogpiled civ even when they're already involved in one war. As a result, it's very likely for one civ to just be absolutely doomed early.

On the flip side though, AI vassals love to break away in KMod. A change in state religion, often precipitated by old religions "dying out" in cities can facilitate this. So while runaway AIs are more common in KMOD, they're also more susceptible to implosion.

 

[AspiringScholar]

Nice analogy from the colloquial use of the word "orthodox" lol.

In all seriousness though, why wouldn't you prefer Karadoc if it just makes the AI better and more challenging without just throwing it a bone? While the puzzles of trying to optimize your own play to outpace the ridiculous bonuses the AI gets on Immortal/Deity are fun on their own, would it not be more fun and interesting to play against the computer when it knows how to maneuver, refrain from suiciding its armies, cancels trade deals once they become unbeneficial to it, etc.? Seems like that one would be a generally liked mod since it doesn't change any of the rules or content of the game.

My first guess would be diplomacy; e.g., not being able to do things like "Okay, now that Mansa is at pleased, I know he can't DoW me, so I can leave my border with him lightly garrisoned."

--

As far as the actual capitulation logic is concerned though, are there any hard and fast rules or major factors one could anticipate? Obviously the AI will only capitulate if it thinks that it's doomed, but that doesn't really explain why it does so swiftly among wars with itself, when (AFAIK) all other factors are equal in cases where it's between it and myself, such as distance and relative army size, as you've mentioned.

 

[Fippy]

My problem with K-Mod was that it can be tedious..fine for one game now and then, but not much more.
While AIs are much more difficult to rush, having huge armies and constantly plotting leads to (very) slow teching.
Backwards AIs that are no real danger and more of an annoyance, so i had to pick deity again..which usually leads to an eternal grindfest.
In the end AIs can never simulate a multiplayer experience, and there are drawbacks if they get too resilient.

 

[AspiringScholar]

Yeah, I can see how the grindfest isn't particularly fun, especially when the AI can usually outproduce you at the highest difficulty levels. That aside though, I find playing against the smarter AI in general to be a lot more engaging and rewarding, knowing that they can rush you early and not behave stupidly with their stacks. I'm not an especially advanced player though.

So, is there no "rule of thumb" for when the AI capitulates, or an XML value which varies from leader to leader? Is it just "keep fighting and check often to see when they're ready to give up"? It would be helpful to know ahead of time if possible, whether they would capitulate under certain circumstances. If that simply can't be known then oh well.

 

[coanda]

If I recall correctly, to make an AI capitulate requires that you have sufficient power and/or land relative to them, that their own power be below the game-wide average, and that you have at least 40 net war success against them (bunch of stuff can contribute various amount of war success; you get like 4 for killing a unit and 10 for capturing a city, and they can earn success for doing the same to you).

But there's enough moving pieces involved that I don't waste time trying to calculate it all out - as you said, just check often to see if they're ready to give up.

It's possible K-Mod changes some of that; I wouldn't know.

 

[AspiringScholar]

My problem with K-Mod was that it can be tedious..fine for one game now and then, but not much more.
While AIs are much more difficult to rush, having huge armies and constantly plotting leads to (very) slow teching.
Backwards AIs that are no real danger and more of an annoyance, so i had to pick deity again..which usually leads to an eternal grindfest.
In the end AIs can never simulate a multiplayer experience, and there are drawbacks if they get too resilient.

If they fall so far behind in tech, wouldn't they be easier to conquer? How is that more tedious than playing on vanilla deity?

 

[drewisfat]

Easier because you have a longer window til they get rifles, tedious because you'll have to kill a lot more units and you're less likely to be able to close out the game with cav, instead dragging it on another 100 turns with nukes.
Basically the best two units in the game cuirs/nukes are even better in KMOD while everything else is worse. If you can peacefully settle enough land to do a lib breakout the game plays out comparably to BTS, maybe even easier if isolated. But if you're attacked or stuck to 3 cities you probably just lose unless you have an OP UU or your neighbor gets dogpiled. KMOD tends to be either unwinnable games or meta games.
Don't be deterred, KMOD is great! It's just not that great for deity players, because you're meant to play a difficulty level down but the gap between IMM and DEI is too big.

 

[Fippy]

I guess cannons could be added to that list (Cuirs, Nukes).