Know your enemy: AI Buildup and reactionism

[Grond]

An important part of strategy, we all know, is knowing your enemies' basis for troop buildup, troop makeup, and location.

So here are my questions, with your answers based on your knowledge gleaned from hours of playing MTW:
1) Does the AI build up armies on your borders to match yours in size?

2) Does the AI place its high valor generals on "trouble" fronts, or are they randomly sprinkled about? Do the high valor generals seem to be in border states, or is there any sort of organization to placement? What about Kings and Princes?

3) Does the AI use any sort of army sizing and build plan? (I.e. does it build specific units, or is it random?) Are the units present reflective of a sort of survival of the fittest kind of process, where combat eats up the the "poor" units (like peasants) leaving only elite or better units?

4) When attacked in a single province, does the AI typically reinforce, withdraw, fight, and what happens with their troop allocations to surrounding provinces? What formula, if any, seems to be determining whether the AI withdraws rather than fights, and what factors are used for determinants?

5) How does your diplomatic state influence the above? That is, if you are in an alliance with princess marriages, will the AI ignore your troop build-ups? If you're neutral, do they build up more? War?

6) Does the Influence level of your King affect troop deployment by opposing factions? If so, how?

Feel free to explain your answers with pictures demonstrating same, and please, wax euphoric if necessary. I'm very interested in the general overall theories about the AI and how it makes determinations for strategy. I've been paying more attention to what the AI does, and how it reacts, but it seems to me it's not nearly as organized as it could be, and most of the time the brilliant moves it might make are pure accident and happenstance.

Thanks!

[Emerald Wolf]

I have seen on many instances where the AI will ignore, when in ware with another AI nation, the opportinity for decisive victory. (bieng the main driving factor in Medieval battle plans) I have tried to base my own battle plans on AI movement and have found it far too erratic. For example in a war I was watching between HRE and France along my northern border, The french had a significant battle force (numbering almost 3 to 1 over the German forces) available to invade Provence. There was no significant German forces anywhere within 2 provinces. Burgandy had a minimal garrison. Now in my mind even an incompetant general would use a 2 to 1 invasion force to rout the forces in Provence and use an invasion of Burgandy to remove the possibility of withdrawal. This sort of double envelopment (writ large) is on the order of basic training for a General (at least since the Roman defeat at Cannae). My battle plan based on such was to mirror French forces 1x to Provence 2x to Burgandy (being a French ally and not fearing retribution form the Pope as the HRE was excommed at the time). Moreover the German disposition was such that a loss in either porvince would have resulted in instant, massive civil war. (IIRC 75% of German Generals, including 2 princes) were at a 1 or less in loyalty. The emperoro would have been hard pressed to remain in power.
Such a battle would have decimated the HRE and brought the Empire crashing down. BUt the French waffled... why? It seems as though the AI is not as aggressive in fighting wars against other AI nations. My wonder is if the AI can be trained in basic strategy.

[Vladimir]

Well for simple tactics just remember that the AI produces junk units so technology is the key. If you really want a challenge just autocalc everything because the limited number of units that can take to the field at one time is a huge advantage to the player. During auto calculate all troops are considered to be in the initial attack. Only thing the computer is programmed to think is build, build, build until a certain maintenance cap is reached leaving it with junk.

[Ludens]

1) Does the AI build up armies on your borders to match yours in size?

Yes. This can lead to cold wars when the A.I. and the human player are trying to match border forces. Large garissons at the border are seen as threatening by the A.I., and will make it refuse diplomatic overtures.

2) Does the AI place its high valor generals on "trouble" fronts, or are they randomly sprinkled about? Do the high valor generals seem to be in border states, or is there any sort of organization to placement? What about Kings and Princes?

The A.I. does use its best generals in important locations, and that includes their princes. The King is used more conservatively in my experience, but he will still fight if it is convenient. However, if there is a risk that he will get caught in a castle, the computer prefers to withdraw him (if the odds are stacked in his disfavour anyway).

3) Does the AI use any sort of army sizing and build plan? (I.e. does it build specific units, or is it random?) Are the units present reflective of a sort of survival of the fittest kind of process, where combat eats up the the "poor" units (like peasants) leaving only elite or better units?

The computer has building preferences for units and buildings, depending on his "personality". It also seems to prefer building more expensive troops over cheaper ones regardless of personality, but since the A.I. is mostly broke this matters little. Peasants are always built in large quantities.

4) When attacked in a single province, does the AI typically reinforce, withdraw, fight, and what happens with their troop allocations to surrounding provinces? What formula, if any, seems to be determining whether the AI withdraws rather than fights, and what factors are used for determinants?

The main determinant for deciding whether to fight or flee is the number of invading troops, but quality of soldiers and general also matter. If these are in the A.I.'s disfavour, he will bolt to the castle or (if his king is at risk) flee the province entirely. However, if there are strong A.I. forces in a neighbouring province, the computer has the option to reinforce. The A.I. can sneak a look at your moves before making its own, though the developers said there is a limit to such cheating behavior. Sending a feint attack (attack and withdraw before battle) to the province containing those forces will prevent reinforcement.

5) How does your diplomatic state influence the above? That is, if you are in an alliance with princess marriages, will the AI ignore your troop build-ups? If you're neutral, do they build up more? War?

Princess mariages matter little in my experience. It is just another type of alliance, and the computer does not care much for them. In Shogun, it depended on the A.I.'s personality whether he would respect alliances, and it might work the same here, but since the factions switch personality regularly, I cannot say this for sure. Troop buildup at frontiers are always looked upon with distrust.

6) Does the Influence level of your King affect troop deployment by opposing factions? If so, how?

Not as far as I know, but it may be that high-influence kings stand less risks of being attacked.

[Grond]

I did some experimentation with force build-ups, and yes, it appears that the reduction of forces on borders causes a reciprocal reduction, even if you are at war. (In this case, the HRE vs. the English, where I have Province / Isle de France / Champagne/ Flanders, I pulled back all the armies except in Flanders, where the covering army for Champagne had to sit. The HRE forces immediately went way down to about a horse and catapult per province, as they had pressing needs elsewhere (in Kiev). Amusingly, I took Moldavia (west side of the Black Sea) from the rebels there, and the HRE keeps invading and backing out. I continue to reinforce Moldavia, and it's now an island of profit and happiness amidst the crummy governments of HRE, Poland, and Hungary. None of them rules worth a damn. My empire, which nets about 15 K Florins per turn, has a nice fat half a million florins in a war chest with which to support the huge forces I'll need to beat the Horde and Spanish.).

Combined with using clergy to spy on the enemy (at least two provinces deep, using experienced inquisitioners in the border states to knock off any generals) plus a good follow-on cadre of 2-3 spies per state to induce instant happiness in the recaltricant population, it made for a pretty good rolling wave.

If I were to replay it, I'd build some follow-on seige equipment and make sure I had a good high-dread governor to jump in to each province; and also do an attack on someone like the Hungarians, with whom I do not trade, so I can rack up my pope warning for them, then beat up the HRE with impunity.

[ichi]

To spring a 'surprise' attack simply build up your stacks in interior provinces. For example if I'm English and want to invade stupid Flanders I build a stack not in Wessex, but in Wales, then move through Wessex and attack before the French 'react'.

ichi

[Ludens]

Another thing is that the A.I. has two systems for invading: the planned invasion (takes years to prepare, this is the one your spies can warn for) and the opportunistic invasion. I think the degree of opportunism is influenced by the A.I. personality, but I am not sure.

The A.I. is aware of the presence of your spies and assassins (or at least: it knows when your assassins target one of their agents), and I understand they are unwilling to sign alliances when you have these agents in their territories. I have never seen this confirmed, however.

[Patron]

The AI deploys troops quite well, but tends to go bankrupt.