• We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.
Well try to delete all missions and see what happens: AI will react pretty much the same way it does with missions, so i really doubt missions have such a big impact on the way AI makes decisions.

And ive never seen an AI attack someone who is bigger, unless they are at war with someone bigger and/or have their armies destroyed/otherwise occupied. (with only exception being countries who are over bad boy limit)
 
Well try to delete all missions and see what happens: AI will react pretty much the same way it does with missions, so i really doubt missions have such a big impact on the way AI makes decisions.

And ive never seen an AI attack someone who is bigger, unless they are at war with someone bigger and/or have their armies destroyed/otherwise occupied. (with only exception being countries who are over bad boy limit)

I played a lot of vanilla EU3, and though I can't say for certain how much of an impact the missions themselves were, when I switched to IN, the behaviors of countries changed a lot.

Never? It happens all the time in my games. Even OPMs will attack a larger neighbor if they've got some good alliances. In fact, provoking smaller nations into attacking me is my favorite way to play! I'm certain I'm not the only one.

These accusations seem rather baseless. I would like to see some proof.
 
AFAIK the 'generic' AI is there so that they don't have to code what each tag should do in each eventuality. This means you can switch the Timurids tag with England and there shouldn't be any change in how the tag would play in comparison to the original.
 
You know what i think would be absolutely fantastic, but a nightmare to code... scaled AI ability based on leader stats. So a monarch with ADM 9 would run a country very well but with MIL 3 would be a bafoon with war. Ofcourse this would mean 7 different levels of AI ability per stat with 3 stats and thus 21 different varying AI difficulty levels... I am sure some dev would wind up in a fetal position under his desk... sigh. What dreams may come.
 
The point to any AI is to make the game fun and challenging, not to satisfy some silly standard of synthetic sanctity.

EU does that, in part, by giving us so many countries to choose from that our level of difficulty can be chosen with precision. Whether the AI "cheats" only matters if you're looking for something other than fun and challenge.
 


Here is the example of England's AI completely failing, in my humble opinion. It attacked Flanders and after that completely ignored Scots who invaded the island, and to my astonishment Scots completely ignored SAVOY who invaded them...

Note that this version is heavily modified (mostly new provinces and stuff) but still same AI applies...

There is definitely something wrong with the way AI treats home vs overseas and/or garrison proportions (if it works anything like HoI2 that is).
 
Civ V AI is regarded as retarded on any difficulty level. You missed the shitstorm on CivFanatics?
Then again, there seems to be raging on any aspect of Civ V. It's not very well received.

People are upset that it isn't Civ IV; they'll complain about anything. I don't listen to those fanboys. That's one of the most difficult game communities out there.

Why does everyone moan about the AI ? its the best AI I've ever seen in a strategy game ! just get ETW and you'll learn to love EU3's AI

also, if you think devs are too lazy to make a good AI, then make one yourself lets see how good you are at it
Play Warcraft III. Its 7 years old, and the AI actually knows how to play the game well!
 
i think the AI should bother to get a MA treaty when one country is separating it in two pieces, i've seen too often Milan or Burgundy defeated like this ...
 
So what if I place 2k infantry next to the enemy, they assault - and yes their 5k cavalry is stronger than my 2k inf - but wait! I have another 6k cav one province further away that you didn't consider and once again the AI is "stupid". I'm not saying that your's or anyone else's suggestion is bad, just pointing out how quickly it can fall apart due to the player doing something else - nicely ilustrated by Premu above.

It might be reasonable to instead code the AI not to go on such sieging rampage, but instead to progress the war in a more steady and systematic way. By this I mean not sieging everything in sight, but only provinces where besieging forces can be immidiately reinforced by the main field army. You know, just like reasonable commanders did IRL.

Another alternative I think of is to dramatically change the supply limit feature. Instead of a constant number, it should be a slowly replenishing resource that armies eat up when they stay in provinces. For example - a province has a max supply of 10, which is replenished by 2 per each summer month (obviously, food and such weren't produced during winters). An army of 4k will eat up the province's resources and start to starve after 3 months, while an army of 10k has only 1 month, just long enough for a fast assault.
 
Last edited:
I think the biggest thing to work on is the peace deal system. If you are soundly defeated, king killed, armies in retreat, provinces occupied.. The AI is more than happy to peace out for 75 ducats. If they continue the war, and I try to hold them off, they end up loosing, with their armies destroyed and provinces occupied.
 
A simple solution to that is to make separate, random AI bots, all using different tactics and gameplay styles. I'm not sure many of you are familiar with games with well programmed AI. In a game with well programmed AI, the higher difficulty setting have a lot of trickier behavior, smart movement, smart positioning, good calculations on the appropriate time to attack.

Really, the AI in this game is extremely linear. If you've only ever played Paradox games, you wouldn't notice. The Ai in these games behaves very rigidly. If only the AI had some flare to their tactics, things would be more interesting.
 
The AI does have different personalities for different countries. If you use the AI viewing console command, and hover over the shield, it displays their current personalities, who they feel are the biggest threat, who are their friends/friendly towards, etc.
 
Already by the power of mighty pause button.

Pause is another EU3 feature unknown to the AI.
Paradox should code the AI so it know how to use the pause bar!
:D
 
AI in EU3 is actually REALLY good. People just like to bitch about it to make themselves look uber. I don't know why these people even bother to play EU3 if it's so easy.
 
"Play Warcraft III. Its 7 years old, and the AI actually knows how to play the game well! "

It's also a real-time streategy where a small number of "sides" have to manage a very limited number of resources in a finite and fairly limited space. Diplomacy is also somewhere between extremely limited and utterly inexistant. And there's pretty much only one goal (in skirmishes or whatever random maps are called). In other words: a game with a very limited scope.

As opposed to a grand strategy game like Europa where over a hundred "sides" each with their own unique situation have not only to fight military campaign, but also constantly evaluate which nation is most likely to be a threat to them, plus to keep tracks of hundreds of game options (decisions, advisors, NIs, leaders, colonization, trade, religion...). The very fact that the AI has to keep track of diplomacy and threats - from among more than a hundred possible source of threats - constantly (as opposed to Warcraft where people are either with you or against you and there's remarkably little variability there) already adds ridiculous levels of complexity that aren't there in Warcraft.

Even if the Europa AI was much more complex than the Warcraft III one (and I wouldn't be surprised to learn it is, given how much more it has to keep track of than a Warcraft AI, but that's just speculation), the fact is its complexity would be spread all over the place trying to match all those features Warcraft doesn't have, instead of being all focused on doing a single thing. Result: the Warcraft AI appears better at what it does.

Plus, again, warcraft has only so many "sides" needing to run AIs at any given time. Europa has something like an order of magnitude more, thus far greater computer demands right there.

Bottom line: expecting a game that has dozens of AI using dozens of distinct game features of which military thinking is only one to have a AI that outperform at military aspects a game that has a handful of AI that have no purpose other than to figure out military strategies is just plain not realistic. Even with a few years difference between the two.
 
Last edited:
Granted, it's possible to conquer Russia starting with Tuscany or Münster, while the AI usually stagnates, when it is not annexed by neighbours, but when you think about it, within a few centuries, Rome went from a single city with powerful and not quite friendly neighbours (they would say 756 BC, if my memory is right) to hegemony over Italy and the Mediterranean by the 2nd century BC, and a EU game starts when some kind of basis for power has been founded, not when your original leader kills his twin. Genghis Khan conquered most of Asia within one lifetime, and he didn't actually rule anything during most of it. You're basically supposed to create a new Rome or a new Mongol Empire in similar games, it's the point of the game. So while the AI can be improved, it's no much more stupid or ineffective than some historical human leaders really were.

However, I don't like to see that it is so easy to control separate pieces of land, even without massive (or even marginal) naval superiority. When the AI faces an ennemy of similar military might, it should be more focused on where a threat exists, and use initiative only on meaningful targets if faced with them. Landing 5k soldiers in Danish-controlled Gotland should not happened when you're losing Stockholm and risk having the rest of your army routed. Also, I'm not really bent on colonizing, so I don't know much about what happens in the New World, but from what I have seen, the AI either has 90% of its army in the Americas, or none at all, same with the navy. With a relatively peaceful Denmark, I conquered most of Great Britain because all the fleet was God knows where, and only a few units where in England, and in another game, I saw several doomstacks around the coast of Spanish South America, representing most of Spain's military, though no war took place, to later walk by in their lands with only 5k cavalry, with no sign of danger. I don't know how the AI chooses where to wage war or to place troops when there are several possible areas, so I don't know how difficult changes would be (though I don't think it's a very easy proposal to tackle on, because several 2k armies in every single island is not much better), but it's the only think I'd like to see that would help a more realistic and difficult experience. It would give a more plausible result to expeditions such as that of Scipio Africanus, either in the place of the Roman Republic, or that of the Carthaginians.

And I also think some part of randomization at some point in the process would be good, though I don't know how it could be implemented easily or dynamically.

About the small fleet attacked with small fleet * 2 problem... If it sees 4 galleys of their coast, and attacks it with all their 15, and you can just send in 50 galleons to destroy them, I think it's better than them losing only 8 to the operation (even though the player can just rinse and repeat until the same result happens). I haven't observed that problem, though.