AI tendencies during War

[Mantic0re]

1. War Bribes: What factors affect the price an AI requires to go to war for you? Obviously their disposition toward the human and target AI, but what else is involved? How much of this can I manipulate to my advantage?


2. Unitprob: I've noticed some AI build more units when they are at war, and others just keep building stupid (Wonders / naval units while at war with land target / missionaries / markets in low commerce cities). I was under the impression the "unitprob" determined how likely an AI was to build military units at peace-time. Is there another for an AI at war? Is there any way to predict which AI will continue to build stupid and which will quickly double the military numbers.


3. Unit choice: What determines the unit choice made by the warring AI when they have the necessary resources. I remember cutting metal to avoid maces/pikes, only to face more fortified longbows. In hindsight I think the metal unit top defenders required less sacrifice on my part. I like how they seem to prefer grenadiers over cannon and how coastal cities can't seem to produce land units. Are there other little tendencies that can be exploited. I realize this question is almost irrelevant if you bring enough siege but long reinforcement chains mean that the last stages of the war are often a little light on siege.

4. War targets: Is there any way to predict which city an AI will attack first if there are two target cities across a long border?

 

[ahcos]

1. Is the target already at war? If yes, with how many ppl? Powerrating? How far away? Does the bribed AI have units, or will it have to build one because it comes out of a war (in other words: the own power rating)?

You actually can manipulate all of that for your advantage except the distance, but ofc it's questionable if it's worth it to raise the powerrating by gifting units or techs.

2. If i remember right there's a number in the reference guide adressing exactly this number. Some AIs raise their unitbuildprob while at war significantly, while others almost build as little as during peace time. Afaik, #1 in the "bastard chart", Ragnar, raises his unitbuildprob by or to 100%, not sure here.

3. Completely random from what i've seen, but as long as an AI has metal or any strategic resource, it'll prefer to build those units instead of plain archer. All AIs will prefer to build their UU.

4. Check:

a) distance to the border
b) distance to the empire overall / distance from capital
c) older cities are more likely to get attacked (but this is solely based on experience and not proven by the gamecode, although i'm pretty sure about that)
d) IF the AI already failed to take a city, it sometimes changes targets as long as there's another city in reach that fulfills the requirements

Resources and strategic things like if the city's on a hill seem to affect the decision too sometimes, but not in any real significance. Distance is really the thing that matters the most.