AI psychology: a thread attempting to analyze it´s behavior (contribs. welcome)

[Noir]

I say that it is a "suspicion" as i can't (and don't have to) convince you for the truth of my conviction and neither can you of yours as your main argument is extensive experience. However i need to respect your opinion, much like you are respecting mine so we can have the very nice conversation we are having about our soft spot Deus.Ret.

For what Cambyses II comments, i'll say that the AI certainly did take account of the maintenance cost of its boats in building his trade network while i was playtesting campaigns in my home mod where the fleets cost apprx. 50-60 flrns to maintain each, in every campaign. I have observed this, as well as the fact that the AI would wait to make a port in a province before deploying a fleet there in order to trade if he has the chance. His invasions were always aimed at provinces that would decrease significantly the maintenance costs he was getting ie in the middle or the other end of the trade "line". His tendency to do so was increased as i mentioned by the amount of money he was getting by the trade; if he was succesful he would definitely invade - first rebels and if not available eventually someone else.

For these i am certain beyond any doubt, i've seen it happening in every campaign i've playtested. I can pass you on the files of my alteration of MedMod IV so you can run test campaigns your self, if you (or anyone else) wishes. Its also easy to observe, because there are few fleets that do not aimlessly hanging around in all seas like in vanilla due to the maintanance cost and because of the basin separation i've told you in another occasion Cambyses II.

If the AI actually does this because the cost is so high, i don't know (and so it does act more at his "free will" because he can afford to in vanilla). However, i can tell you that even if the latter is true, it proves that the AI is capable of good play consciously to a certain extent on his way to become huge.

Once he gets huge he overextends and blows off by himself and i believe this is hardcoded.

For the AI invading for economic reasons - they are not pure i agree, however they seem to be included in the equation, when doing invasions, and also when Crusading. It is incredible that Egypt or Algeria are the targets of Crusades 95% of the time in my games - as they are the more lucrative ecomonically provinces in muslim lands.

The factions you mention stagnate because the AI factions find it risky to invade - its not like they can afford to lose as they will be anhihilated. Surprisingly, enough the AI does care to stay in the game, and as i said earlier factions that are beaten severely and "sense" anhihilation, will sue for peace. I've seen that happening and read about it many times. It is also happening in the SWs mod, that almost every faction is in the state of the Aragonese, including the player. The AI personality of them in the mod is set to "expansionist" for all. They take ages to take off though, because they are cautious as they are all surrounded by enemies. Things may go wrong, and the Ai aknowledges that. Apparently it has a certain threashold embedded for invading and until that is reached he won't do it or he will invade with very few troops - this is mentioned in the files as "opportunistic attack" ie its not a full scale effort but more on the basis of "see if we can exploit that gap".

The Serbs, a similar case as the ones you mention in my home mod of MedMod IV, wait for the right conditions around them to invade. Where the game is wrong is in the buidling costs and building flow as well as the bodyguard maintanance costs that prevent the AI to be close to the green, economically. If the military buildings didn't take an age and cost a fortune for the AI to build as he goes on to complete the forts and then start by building them all one by one, you'd see that when he does makes a move from the 1 province faction state, he can actually come on top if not bunkrupt.

"Idiotic" attacks from the part of the AI, as it was mentioned by The Unknown Guy and Cambyses II, mean economic health and also thus, the chance to get the newer units. The AI initially, especially in vanilla,spams armies of very low tech troops, in order to deter others from invading. However these cost him dearly. For the player its easy: slowly disband and then make the better ones, but the AI can't do that. He has to get rid of them by making war. So he invades with inferior troops, raising the palyer's eyebrows in the mean time. However for him this is the way to progress, as he can have positive net again and so build the newer/better units. This is why i haven't touched vanilla for a long time now, as the difference between the units in eras is huge - essentially a battle winner, but the AI doen't have a chance in keeping up building them like the player does hence the hardcoding of building all military upgrdades first. This means that he neglects the economy ending up in the red for most faction apart from the fortunate few that are destined to become huge so the player has something to do.

All in all its simply bad design, that was introduce to make the campaign "deeper" also read "more conventional RTS". Now of course after RTW, it seems like a "harmless vice".

The game's tech/build tree accounts for the player only but not how the AI does things, and this makes him look stupid. Of course he is, but i think much less than most of us think. What the game really needs to become challenging throughout its length, is a smoothing of the spending flow and income of the AI factions and also ensurance that this flow remains constant together with balanced rosters. For example the Ai doesn't use hybrid units effectively - he never uses them in melee. The very fact that they exist is a spoiler as they are there just for the player.

Noir

PS Apologies for playing the "odd" man in this, it happened unintentionally.