A Grand List of AI Problems with the hope they will be fixed

[Slyspy]

Alternatively make badboy limit be inversely proportional to size.
If you are very large you will have very little room to maneuver before other countries will get concerned, but if you are small, they will be less concerned.

This is a genius idea. Whether it solves the AI problems or not this is how BB should have been implemented IMO.

 

[ulmont]

Alternatively make badboy limit be inversely proportional to size.
If you are very large you will have very little room to maneuver before other countries will get concerned, but if you are small, they will be less concerned.

But remember that the Castile-Aragon inheritance, Castile / Spain overrunning the South American pagans, and the Ottoman Empire's expansion should still be allowable...

 

[unmerged(503)]

well said, so to speak. your points all lead me to believe, the game was designed with multiplayer fully in mind first, and single player second, its the only logical conclusion I can arrive at based on the game as it stands, it appears what we are waiting for in single player, a much more agressive and interesting AI, was in reality intended to be the multiplayer arena at work, IE the players are the "agressive AI"...the rest of the map is simply a passive backdrop awaiting the actions of the players, not the AI. this is really the only conclusion I can arrive at. at any rate im bored with the game, time to shuffle off and focus on something else.

You can not see the distinction in arguing that the AI behaves too aggressively towards the minor nations (especially minor v. minor wars), while behaving too passively towards majors (unless they are one province minors, in which case they are quite suicidal, for whatever reason)? I cannot fathom how anyone could percieve that position as being inconsistent.

But you have pointed one thing that is quite correct. After a time it becomes fairly useless to attempt to argue for reform. Every thread twists into the same chasm. A "problem" is pointed out. Several people agree saying they have noticed that problem too. Then a number of people post saying that isn't a problem at all, but the most wonderful design feature ever concieved. Then more people post about how to compromise solutions to the "problem" that keep "feature" aspects intact. Debate ensues. Then, by the end, the entire situation is so muddied that the perception has been created that the problem really wasn't a problem at all, but merely a feature that was interpreted positively by some and negatively by others.

Larger AI powers will not go to war with other, larger powers under almost any circumstance. They break alliances to avoid wars. They sign white peaces immediately if you "force" them into wars by loading savegames as different nations or modifying the savegame text file. They simply won't fight great power wars. The majors preference is to gobble up minor powers and then sit and do nothing if no more minors are accessible. Minor powers on the other hand are suicidal and aggressive. They attack each other and then, inexplicably, they attack majors. Soon they are all gone. The longer you play, the more static the map, until, finally, there is nothing to do. Is anyone observing a different pattern than this? Is anyone seeing something else?

Some feel that this is an expression of the brilliance of the AI. Now it evaluates its chances accurately and if it isn't overwhelmingly likely to win a war, it won't engage in it. Hurrah for the AI! No longer will the player will able to take advantage of France after she exhausts herself fighting a great power war against Spain to liberate Franche Comte and Rouisillon. The AI is wonderful because now it is a challenge for world conquest players and is less historically predictable.

Except, it doesn't do anything, right? All this new static AI does is form stationary roadblocks to world conquest players blob-like expansion. It doesn't exhaust itself fighting other forms of itself, but it doesn't do anything either. It won't attack strong players (though, apparently, it might attack medium players). And it gives the player nothing to do, but sit and plan his next war.

Is that what people want from EU3? To go from war to war gradually conquering the map? Didn't people kind of enjoy just sitting in the game, watching the world develop? Isn't there any roleplaying aspect at all left in EU3's fandom? Don't you want to feel that you're in a realistic environment where large nations are rivals to each other, not speedbumps to slow the player and create an artificial "challenge"?

But, anyway, as I said, the waters are muddied. There is no "problem" here. There is just a difference in opinion. Those of us who want something more, we're just bitter historical gamers who want our old AI files back and can't embrace this new, wonderfully dynamic AI (who, ironically, is not dynamic at all and is far more predictable than any EU2 ai-file run AI).

I think EU3 is a dead-end game because of the AI. I don't think it will be changed to more "realistic". I think the design strategy IS working as intended and that the AI is designed to balance itself as blobs to prevent easy player expansion. I think all the mods in the world can't fix this problem. What baffles me though is how the defenders of this newly enhanced AI aren't incredibly bored by 1680, 1690, 1700, etc. watching the AI do absolutely nothing. But, I wish those who carry on this discussion good luck. Put me down as being in favor of hard-coded tweaks to the AI's routines that allow it to consider fighting major v. major wars to give the game flavor, to keep the map from ossifying, and to just make the game believable, but put me also down in the column that thinks, based on betatesters and moderator comments, that things are working exactly as Paradox intended.

 

[Antoine]

This game is a big disappointment for me. On paper the "dynamic" system was supposed to create better "what if" situations than the old "determinist" system. But instead of coding a dynamic AI trying to simulate how nations of this era would have reacted to different situations, they have coded a very poor gamey AI, only considering the possibility of short term easy territorial gains and self preservartion, making the game totally unable to create something looking like a plausible alternative history, and worse making it a very boring experience.

 

[Fawr]

I'm disapointed.

EUIII was supposed to break the "England will do X in year 1623" pattern from earlier games. However it was supposed to replace it with "anyone in a simlar situation to England in 1623 has a decent chance of taking action X". So England may not colonise North America, but its likely some similar country would.

Hearing that wars between two large countries is designed not to happen, despite the many historical precidents is not what I want to play. This quote in particular is what turned me off

the AI is NOT HISTORICAL. it will not make historical mistakes, it will do whatever it thinks is best for itself. Major wars against major powers that it thinks it cannot win are not. period.

Not only does it not understand the whole problem (Two countries of roughly equal size still has one country that would benefit from a war), but it shows how EUIII's ambitious grand plan has been poorly executed...

 

[Slyspy]

But remember that the Castile-Aragon inheritance, Castile / Spain overrunning the South American pagans, and the Ottoman Empire's expansion should still be allowable...

Those events didn't exactly make those countries more popular though did they? Just that no one was stong enough to go on the offensive against them (Armadas and Vienna aside).

 

[unmerged(12921)]

I'm disapointed.

EUIII was supposed to break the "England will do X in year 1623" pattern from earlier games. However it was supposed to replace it with "anyone in a simlar situation to England in 1623 has a decent chance of taking action X". So England may not colonise North America, but its likely some similar country would.

Hearing that wars between two large countries is designed not to happen, despite the many historical precidents is not what I want to play. This quote in particular is what turned me off. Not only does it not understand the whole problem (Two countries of roughly equal size still has one country that would benefit from a war), but it shows how EUIII's ambitious grand plan has been poorly executed...

The sad truth is that the designers got it dead wrong. Wars do not generally occur when one side is noticably weaker than the other. The lessor side either "gives in" to whatever demands are being made, or goes out and finds itself a "bodyguard" (an alliance large enough to "equalize" the odds). There are exceptions, but as a general rule common sense and realism prevail.

Wars arise when both sides feel confident in their strength, position, and/or military prowess. One side or the other is eventually proved wrong in it's assumptions---but virtually never does a nation start a War it has no hope of winning, and only seldomly does the weaker side in a dispute chose combat over negotiation. The AI's whole programming keeps creating "inverted" results.

 

[unmerged(12921)]

Shameless BUMP. Too many good ideas and solid problem ID's to just be alowed to drift away....

 

[ConcordantNexus]

3 ideas: BB, negative stab lowering perceived strength & shuffling the 8 lucky nation

Can we piggy-back that idea about scalable BB to Stability somehow? Stability costs are already scaled to size.

----

Also, is there a way to factor in a negative stability penalty to strength assessment by neighouring powers?

What I mean by this is: if France is wracked by revolts from the Catholic-Huguenot conflict and at low stability - shouldn't the neighbouring AIs decide that it is weak and attack?

----

Alternatively - the balanced of powers stalemate thing is probably caused by the lucky 8 nations.

Perhaps someone could write an event that fires every 20 years that deals out the 8 lucky slots to 8 totally NEW nations?

The previous lucky nations should be big and no longer pacified by the lucky-we're-saving-ourselves-for-when-the-human-attacks-us-in-the-endgame.

The new lucky nations just *might* be strong enough to hold off the previous lucky nations' unwanted attentions - and would possibly get big enough to set up the next round of dynamic wars which will hand the lucky torch over?

 

[ConcordantNexus]

Forts, Ships & Armies interaction for BB

Don't have EU3 yet (ordered last week, expect to receive it this week).

How about making BB more brutal?

Armies = offensive
build 1 regiment = 1 BB
hire 1 mercenary regiment = 1 BB
disband 1 regiment = -1 BB

Forts and Navies = defensive
Each level of fort built = -0.5 BB
Heavy ship = -1 BB
Medium ship = -0.50 BB
Galley/light ship = -0.25 BB
Capture a heavy ship = +0.50 BB
Capture a medium ship = +0.25 BB
Capture a galley = +0.10 BB

In effect, this would function a bit like the reverse of Victoria's prestige

----------------------------------
Also, since this period was one which saw Europe's rise, shouldn't BB reduction be fastest for

Latin>eastern>muslim>chinese>pagan ?

 

[bovinespy]

The sad truth is that the designers got it dead wrong. Wars do not generally occur when one side is noticably weaker than the other. The lessor side either "gives in" to whatever demands are being made, or goes out and finds itself a "bodyguard" (an alliance large enough to "equalize" the odds). There are exceptions, but as a general rule common sense and realism prevail.

Wars arise when both sides feel confident in their strength, position, and/or military prowess. One side or the other is eventually proved wrong in it's assumptions---but virtually never does a nation start a War it has no hope of winning, and only seldomly does the weaker side in a dispute chose combat over negotiation. The AI's whole programming keeps creating "inverted" results.

Whoa - I'm not much for the "EU3 sucks!" threads, but this makes perfect sense to the amateur historian in me. Of course, there were rebellions and such, but as a general rule, from what I know of the EU time period, what you say rings true to me. IIRC, many massive wars were fought amongst the major powers of Europe for minimal territorial (or other) gains. This should be the norm in EU3, as well. That, combined with a greatly reduced badboy limit would by themselves go a long way towards modeling the general features of geopolitical reality back then.

 

[Duuk]

On a side note, add me to the list of people that hates the fact that France will never fully form, Spain doesn't form, and the great powers love each other and never fight.

 

[unmerged(25298)]

A new idea?

I think the answer to more major v. major wars is to introduce more reasons for warfare and more "intermediate" victory options. Another is to have the AI favor intermediate victory options.

I don't know how difficult to program these suggestions might be but how about:

1) More Succession Wars and PUs.

2) Force Merchant Non-Compete (i.e. two merchant republics duel it out - winner is immune to competition from the loser in CoT).

3) Force Monopoly - Winner gets an automatic monopoly in loser's CoT.

4) Hastier trigger for BB wars the more disproportionately large the BB is (sorry world conquest players...)

5) More periodic nationalism revolts later in the game depending not just on stability but also prestige and innovative/narrowminded, serfs/free sliders, prestige/fine arts academies/universities.

6) Indemnity - loser pays % of monthly income for X number of months....great way to punish econ. only nations.

7) More bonuses for HRE + Papacy controllers and then 'influence' or 'grab the crown/mitre' wars. This leads to AI actually giving 2 pennies about keeping the HRE intact and models historical pope/anti-pope, etc.

8) Model the spain/protugal treaty - loser gets shafted by colonial demarcation option - i.e. lose all provinces (or don't get to colonize empty provinces) beyond some preset natrual boundries in game files - "east of mississippi", "beyond the amazon", "north of the rio grande" "caribbean islands" etc. Can be revoked only by DoWing the winner of the treaty...

9) Certain province flags --> immense hostility...austria should flip out if France penetrates into Italy beyond the savoie, as should castille/aragon/spain. Stops lots o' blobing.

10) Unite Christendom option - immediate DoW on all non-vassals in Europe, certain key province triggers, certain massive upheaval events (in uniting nation and others), could have happened historically, triggers depending on size, religious unity, deus vult national ideas, sliders, etc.


I WOULD LOVE SOME FEEDBACK THANKS GUYS

 

[unmerged(12921)]

"1) More Succession Wars and PUs."


Can't say I would support this. PU's are already far too common in the game now as compared to reality.

 

[bovinespy]

I think the answer to more major v. major wars is to introduce more reasons for warfare and more "intermediate" victory options. Another is to have the AI favor intermediate victory options.

I don't know how difficult to program these suggestions might be but how about:

1) More Succession Wars and PUs.

2) Force Merchant Non-Compete (i.e. two merchant republics duel it out - winner is immune to competition from the loser in CoT).

3) Force Monopoly - Winner gets an automatic monopoly in loser's CoT.

4) Hastier trigger for BB wars the more disproportionately large the BB is (sorry world conquest players...)

5) More periodic nationalism revolts later in the game depending not just on stability but also prestige and innovative/narrowminded, serfs/free sliders, prestige/fine arts academies/universities.

6) Indemnity - loser pays % of monthly income for X number of months....great way to punish econ. only nations.

7) More bonuses for HRE + Papacy controllers and then 'influence' or 'grab the crown/mitre' wars. This leads to AI actually giving 2 pennies about keeping the HRE intact and models historical pope/anti-pope, etc.

8) Model the spain/protugal treaty - loser gets shafted by colonial demarcation option - i.e. lose all provinces (or don't get to colonize empty provinces) beyond some preset natrual boundries in game files - "east of mississippi", "beyond the amazon", "north of the rio grande" "caribbean islands" etc. Can be revoked only by DoWing the winner of the treaty...

9) Certain province flags --> immense hostility...austria should flip out if France penetrates into Italy beyond the savoie, as should castille/aragon/spain. Stops lots o' blobing.

10) Unite Christendom option - immediate DoW on all non-vassals in Europe, certain key province triggers, certain massive upheaval events (in uniting nation and others), could have happened historically, triggers depending on size, religious unity, deus vult national ideas, sliders, etc.


I WOULD LOVE SOME FEEDBACK THANKS GUYS

I really like #2, #3, and #6. The others look OK, too - I just really like the idea of having more economic possibilities for victory, as opposed to the standard territorial ones.

 

[JScott991]

Sorry to bump this thread, but just some evidence of a gameplay situation where the no major v. major wars really contributes to causing a game to "end" early.

Note the following screenshot of a game I am abandoning to start over.



France is nearly completely walled in by Spain and Austria (I'm playing Austria). The current AI will never go to war with Spain or Austria, so what does that mean? It should mean that France would do everything in its power to destroy the Habsburgs through competent alliances with nations such as Turkey or Protestant Germany and through taking advantage of their interior lines to defeat an over-extended Spain. But not in EU3! In EU3 it means trickling expansion of course! You can see where France has already annexed Lorraine. In the current game year, they are at war with the Palatine. Why? Because the Palatine is the only small minor that borders them and the AI majors will only gobble up minors.

This is despite the fact that Spain owns numerous French core provinces, has an army tech level 3-4 levels lower than France (I can't explain this, except to say that the AI always has the trade tech bar maxed out when I load Spain up to check on what they are doing) and is constantly fighting off revolts.

But the AI will never even consider going to war with Spain as France. So, with no Franco-Spanish War possible, France will try to trickle into the HRE or Italy, producing ludicrous boundaries. The only way to stop this, as Austria, is for me to go to war with HRE minors and vassalize them. Austria has to actually attack Catholic states with +100 relations, in order to stop insane French blobbification that would produce borders too hideous to look at (can you imagine, in this map, France owning the Palatine, Hesse, Cologne, etc.; it would be ridiculous). This game is dead. There is no point to continuing to play it, since there is no logical way to explain a further two centuries of French inactivity (except to join Scotland in constant DoW's on Leinster, followed by white peaces several months later when neither nation can figure out how to use their transports to invade Ireland). I still can't understand why we couldn't get a consensus outcry on this issue, but I guess people aren't as frustrated by absurdity as I am.

Please, please, please Paradox fix this.

 

[unmerged(43915)]

My thoughts on AI

My thoughts on the AI tend towards those of the Original Poster. I want the AI to be active and do stuff, and as it is, it runs out of things to do after a while. I want the AI to recognize threats and move to counter them while it still has a chance, but as it is it just sits there and does nothing. I want the AI to have some reasoning behind the way it makes alliances, and to honor those alliances once made. Finally, I want the AI to be really strong, to play with a more generous sense of self-interest in mind, so that it can both prosper on its own, and kick the crap out of me when I, as a player, do something stupid. If that means throttling my growing power as an infant, than so be it - it will make growing into a power a real challenge, as opposed to the mildly interesting diversion it currently is. I like challenges.

I have a series of ideas for how this might be fixable. I dont' know how the AI is coded, so I don't have any idea for how feasible this might be, but I figure I'll post it anyway.

To start with, the responsiveness of the AI to certain stimuli needs to be enhanced. Right now, it ignores these three important game aspects, resulting in annoying, silly, and self-destructive behavior.

First off, the AI needs to be coded to honor its alliances better. Given its current habit of dishonoring any call to arms against a power bigger than four provinces, or for that matter, declaring war on its allies if they seem to be losing, nothing can be done. In and of itself this will do little, but it is necessary if my later suggestions are to be possible.

Two, the AI needs to actually respond to badboy. As it is, only really small countries care about badboy. Big powers will never declare war on you for badboy if you are a significant power yourself. This is broken, and is a large portion of the widely observed "suicidal minor" issue. Given that badboy is BY DESIGN a silly and non-rational mechanism designed to force AI's to counter threats, arguments about AI rationality and self-interest hold no water when it comes to this point. As it stands, lowering the badboy threshold in the defines file has little effect on how the AI's behave (aside from provoking suicidal OPM's to declare war), and in particular how they behave toward each other.

Three, the AI needs to value stability a bit more. Right now, you cannot use stability modifications as a stick to make it change its behavior, and so it is happy to take a -6 stability hit in order to break an alliance and a royal marriage and military access and good relations to declare war on it's former friend.

Fourth, if the AI is going to attack a country for war exhaustion or badboy or whatever, it really should do that WHILE THAT COUNTRY IS STILL AT WAR. Right now, it always seems to wait for the target to END its long and draining war before declaring war. From a purely self-interested point of view, this is rather stupid, as it allows the target nation to move its troops to the new front without worrying about breaking off seiges or two fronts or anything. The EU2 AI was much, MUCH better about this. The fact that it is almost always OPM's that declare war in this manner does not help, either.

That fixed, I would suggest a system to encourage it to form alliances that make sense. Once made, these alliances would at least be a start towards simulating a balance of power situation, and would encourage threatened parties to act together to counter said threat.

I presume the AI routinely does checks against its neighbors, comparing their strength with its own strength. I would recommend that when these checks start coming up badly, that is, when a particular neighbor is obviously stronger
, the AI sets a durable flag - "I'm scared of X". The worse the power disparity, the stronger/the more flags get set.

Presumably, at any point, multiple nations might well have the "I'm scared of X" flag in action. Alliances should form between those nations. They are banding together against a common foe. It's historically valid, and it makes sense from a self-interest perspective.

However, large nations may never be terribly afraid of other large nations, and so they might never be drawn into such alliances. To goad them, I would have a second variable, anger. Anger flags might be set by things such as declaring war on nations with which one has good relations, annexing OPM's, owning cores, breaking guarantees, having close borders, having colonies near one's own colonies, etc. It would sort of be like BadBoy - but not, because BadBoy is supposed to make AI's declare war irregardless of their strength or standing.

Nations that are Angry at X would be willing to enter into alliances with nations that are Afraid of X, forming large Anti-X coalitions. With the adjusted AI, they would then honor those alliances, and have a good chance of containing X. I would then re-introduce the penalties for dishonoring Calls to Arms, which the AI will now care about because it now pays attention to stability.

These alliances would wait until either they grew big enough that, together, they could crush X, or until one party's Anger meter grew high enough to cause a DoW, or one party was sufficiently weakend by war with an outside source/internal rebellions/bad events/whatever, or until some event causes war to break out anyway, like a Succession Dispute or whatnot. Then you would have Grand Coalition wars between Major Powers that would make sense, and in which all sides would be following something approximating their self-interest. Things would be happening, growing powers would be contained, players would be challenged, the map would not be static, and everyone would be happy.

 

[unmerged(70700)]

I presume the AI routinely does checks against its neighbors, comparing their strength with its own strength. I would recommend that when these checks start coming up badly, that is, when a particular neighbor is obviously stronger
, the AI sets a durable flag - "I'm scared of X". The worse the power disparity, the stronger/the more flags get set.

I love this idea, and X could try to counter the "I'm scared of X" modifier by gifts, selling a core province to the scared nation, etc. Maybe once one of the nations allied against X reaches a -200 modifier, he DOWs X thus triggering the larger wars.
All in all, there does need to be something like this in the game. With the present AI there should be no minors in germany at the start of the game because concievably they'd of been gobbled up long before the game's starting time frame.
Besides, in down times between wars I like to see other nations going at it! Makes the world feel dynamic and alive.

that would produce borders too hideous to look at

I haven't played enough to be sure, but are all the tendrilly, long line, tenticled borders of some countries bothering anyone else? Is this what you mean JScott? It is aesthetically unpleasing, but I guess it 'could' happen sometimes. In a game I'm playing now, Austria stretches from Ottomans to Flanders in a weird multi-headed worm... and Poland is wormy too. I'm not playing either, I'm playing Urbino.

 

[unmerged(14180)]

What happens in EU2? If someone gets too powerful, various nations next door (Or even further) create a coalition and deal with the nasty empire.

I am not seeing this in EU3 as yet.

Even the alliance system is totally silly in my opinion. I can maybe manage to get one ally, and then the rest become 'impossible'.

Theres some changin needed in this game.

 

[pdubz]

I have a series of ideas for how this might be fixable. I dont' know how the AI is coded, so I don't have any idea for how feasible this might be, but I figure I'll post it anyway.

I really like your idea, Zwackus. I would be surprised, but very happy, if they implement it.