The AI's frequently do this. What's the deal with that? It just makes no sense for the Russians to abandon Moscow and go off to siege down Japanese Siberian holdings.
- 13
- 1
AI abandoning its own country and sends troops off to siege down far away, unimportant land.
- Thread starter Vampiresoap
- Start date
-
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.
The crude overgeneralization is that the AI tries to avoid engaging "scary" concentrations of enemy force, while still seeking to lay siege to enemy forts.
Unfortunately, the tuning factors for this make the AI look like a panicky idiot.
Unfortunately, the tuning factors for this make the AI look like a panicky idiot.
- 10
- 8
Try saving and reloading. But with so little info other than a generic complaint it is hard to make a judgement. Who were they at war with, what was their manpower, what year was it, .....?
- 9
- 2
If you play the game enough, you would know what I'm talking about.
- 17
This is a pretty big problem for large countries, not so much for smaller ones like in Europe. Unfortunately, it seems like something we’ll have to deal with until EU5 comes out
The problem is probably not the AI. It is doing probably something sensible considering the game rules. I think it would be much better to consider which game rules are causing this to be a worthy tactic? Maybe losing provinces should cause more war exaustion?
To be honest, there is at least 1 real history example in the time period: Fredrick the Great, literally abandoned every thing he had to invade and conquer Silesia. To the point that yes, indeed Berlin briefly fell to (I think hungarian?) cavalry in a surprise attack...
His bet was that Silesia was stronger economically then the rest of his country, so even if he lost all the rest he would continue to fight in a "Fortress Silesia"... and he won (despite multiple times almost losing)....
But yes, this was an exception. Usually people didn't leave their country unguarded.
To be honest, there is at least 1 real history example in the time period: Fredrick the Great, literally abandoned every thing he had to invade and conquer Silesia. To the point that yes, indeed Berlin briefly fell to (I think hungarian?) cavalry in a surprise attack...
His bet was that Silesia was stronger economically then the rest of his country, so even if he lost all the rest he would continue to fight in a "Fortress Silesia"... and he won (despite multiple times almost losing)....
But yes, this was an exception. Usually people didn't leave their country unguarded.
- 5
- 2
- 2
Well, I have well over 1000 hours into EU 4 plus 20 years of coding experience, so......
I am not saying the ai doesn't occasionally do something that sounds crazy to a player. Usually the issue is that the ai sets a target and sends the army to attack based upon the info and conditions it sees at that time. The ai doesn't reset until the troops get to the target. It isn't really dynamic or capable of reacting to constantly changing conditions like a human. If that was coded the ai would just spin troops around the map doing nothing, not to mention the lag that would cause.
This is why I say try saving and reloading. That causes the engine to recalculate all strategy routines and pathing based upon what it sees at the save. If that doesn't work then the ai has calculated that doing whatever it is doing is best. Even if it makes no sense to the player. It may be trying to reduce risk, to complete a mission, to knock out an opposing ally, etc etc. Or it has just decided staying and fighting is suicide. I have to say rarely in 1.30.4 do I see the ai going tilt.
- 8
- 1
This is probably the 100th thread about this is the last 3 years. There's another on the front page right now. It became known colloquially as the AI fleeing to Siberia. My favorite one was the new player that picked Muscovy and was asking where the AI armies were as he conquered all the steppe land. He said he saw Uzbek hiding in uncolonized land once and never saw another army asking if it was normal. Game just breaks if you pick a major.
- 4
And yet I am playing a game as Oirat with Ming trying to attack me from all quarters when I am at war with them, Timurids and their allies pouring up at me through mountain passes trying to get around my screens when I fight them, and Uzbek/ Nogai/ Great Horde/ Delhi alliance doing everything they can to either protect their forts or get into my interior when I fight them. I have had to reload one time when Ming disappeared on me, turns out they were consolidating retreating troops in the south with new troops and mercs they bought.
I know this goes against the popular ai is trash meme, but the ai really is a lot better in 1.30.4. You just need to help it occasionally with a reload, but that is consistent with the Clausewitz engine. HOI4 suffers from the same issue.
- 2
Yeah probably because 300k Ming soldiers aren't scared of your 30k stack. Try having 100k yourself and attacking an AI that has 30k. Hell, he could have 3 other allies with 20k each and they probably won't group up properly.
- 4
- 3
Meglok dude still doesn't get what's going on or what's causing it lol...And yes, this issue comes up when you are the one with 300k coming up against an AI that has 30k troops. Not the other way around. That example is quite exaggerated too. I think the AI's still run away even when you're just slightly stronger than them. Like 100k vs. 80k.
- 6
I was on the other side of that experience yesterday as Sligo. England and Scotland were carpet-sieging me while my French allies occupied basically all of England.
Edit: I should add that I had 20 War Exhaustion, was facing bankruptcy and would have gone bankrupt if I hadn't debased my currency, so England probably expected me to make a separate peace. My theory was that if I held on long enough without making a separate peace France would win the war and I would avoid losing territory.
Edit: I should add that I had 20 War Exhaustion, was facing bankruptcy and would have gone bankrupt if I hadn't debased my currency, so England probably expected me to make a separate peace. My theory was that if I held on long enough without making a separate peace France would win the war and I would avoid losing territory.
Last edited:
And what exactly would you do if outnumbered 30k vs 100k? Assuming, of course, that your enemy is not utterly inferior at warfare or tech/ideas.
I'm definitely mediocre at EU4 warfare, but seeking battles against a stronger foe doesn't seem like a promising strategy. Siege racing and harrassing to provoke a milder peace treaty where my armies and manpower aren't wiped out seems just better to me in the long run. Or concede a minor loss near-instantly. Since the AI won't to the latter, and pretty much attempts the former, I'd like to hear what you want it to do differently (rather than just better)
- 2
- 1
In my own limited experience, the AI in 1.30 is better than 1.29 and 1.28. I had stopped using the "declare war" button for CNs, because I have had many cases where my Mexico CN with 50k men will gather them in one stack and take one provinces after the other while in the same time 2 or 3 native armies will take 3, and finish to lose the war. Now the AI seems able to calibrate the size of armies, so a CN will no more gather 50k together if its enemies have small armies, and they win this kind of war everytime. The AI is also able to carpet siege (not to the extent to what a player will do, but still far better than before).
I have also not spotted AI armies sitting with 30k men in Malvinas or similir tiny island for decades, as I did in former versions.
Worst side of AI for me is that it won't push for peace sometimes when it is very important. Example is the League war in my last game, where the protestant league main members were three of the four Catholic major powers (France, Commonwealth and Portugal) while on the Catholic side the main members were Spain (Catholic, me), the Ottomans (sunni) and Russia (orthodox). Well in the first years I crushed Portugal (with a 85% warscore) while the Ottos and Russia concentrate on the Commonwealth and took about 2/3 of it, but the Emperor (Bohemia) did not ask any for peace and France at the time conquered all the Germans minor and Bohemia itself unopposed. If the AI had eliminated one of these two countries from the war, we could have won. It didn't until far too late, and in the end we lost (even if Spain itself won 4 Portuguese provinces and lost no land).
Another example I saw in 1.29 was Sweden fighting Russia with each side invading the other country. Sweden had a lead of 40 or 50% warscore but had now only half of the enemy troops and its capital was besieged and at more than 50% chance of falling. A player will very probably have chosen peace in good termes then, the AI did not, lose its capital and the war some years later.
I have also not spotted AI armies sitting with 30k men in Malvinas or similir tiny island for decades, as I did in former versions.
Worst side of AI for me is that it won't push for peace sometimes when it is very important. Example is the League war in my last game, where the protestant league main members were three of the four Catholic major powers (France, Commonwealth and Portugal) while on the Catholic side the main members were Spain (Catholic, me), the Ottomans (sunni) and Russia (orthodox). Well in the first years I crushed Portugal (with a 85% warscore) while the Ottos and Russia concentrate on the Commonwealth and took about 2/3 of it, but the Emperor (Bohemia) did not ask any for peace and France at the time conquered all the Germans minor and Bohemia itself unopposed. If the AI had eliminated one of these two countries from the war, we could have won. It didn't until far too late, and in the end we lost (even if Spain itself won 4 Portuguese provinces and lost no land).
Another example I saw in 1.29 was Sweden fighting Russia with each side invading the other country. Sweden had a lead of 40 or 50% warscore but had now only half of the enemy troops and its capital was besieged and at more than 50% chance of falling. A player will very probably have chosen peace in good termes then, the AI did not, lose its capital and the war some years later.
- 1
- 1
I've seen hannover siege down colonial land in south america during the religious league war. they were an opm.
- 1
Personally I would try to abuse forts and snipe stacks. If that isn't an option I would let them day one peace me out. You're not going to be able to baserace the AI in that scenario lol
I have seen this argument constantly whenever I complain about this. That this baseracing is the strongest move the AI can make in this situation.
My response is
a) no lol, it's not
b) it's extremely unfun
I don't understand the idea that "the AI should make the war as unfun and tedious as possible in order to make the player peace out earlier". This is a game, not a masochism simulator. The AI should be FUN to fight, above all else. Arguing that the AI should employ weird meta-game things like testing the player's patience to make them wanna quit playing is just... bizzare to me. After you reach a sufficient size, 99% of your casualties in a war will come from attrition from sieges, because the AI will do everything in it's power to run away from your troops. This simply isn't fun, it just makes war tedious.
- 9
- 2
- 1
.png){kind=avatar}
AI army movement logic is a very chaotic system by design, AI is constantly on offence and he can only react to homeland occupation at best. AI is part of the game so he sees balance of forces, if the balance of forces is too dangerous AI just can't path in this direction. It's especially visible in CK3, because when this situation occurs in CK3 AI just jumps on the boats. Boat rout is never blocked in evaluations.
What is extremely disappointing is that, TW games had more complex battle AI almost from their inception. Because the AI needs to move units in 3d plain, AI just needs to be more complex. And pdx games are essentially a grid or tiles. I seriously doubt that very limited AI with planning capability, would be worse performance wise than the current chaotic system.
What is extremely disappointing is that, TW games had more complex battle AI almost from their inception. Because the AI needs to move units in 3d plain, AI just needs to be more complex. And pdx games are essentially a grid or tiles. I seriously doubt that very limited AI with planning capability, would be worse performance wise than the current chaotic system.
Last edited:
In a recent war Russia (me) + Georgia vs Ottomans, the Ottomans had 30k sieging Trebizont + 15k in their own territory in the province just next to Trebizont. To me the extra stack was clearly just waiting to join the battle if I attacked the main Otto force.
How do you snipe that?
And how did I answer? Well I besieged Ankara & Smyrna... no point in fighting ottoman invincible troops....
It is very unrealistic, but game-wise, it is a smart move. Why waste all your man in a hopeless battle?
Besides this behavior makes wars significantly easier for the human player IMHO. In another war Russia (me) vs Horde + Nogai, well, they all went siege Kazan & Moscow, in the meantime I besieged everything they had, so I was able to peace out with 100% warscore and minimal casualties. If they would turtle up in a doomstack parked in their capital, I would be forced to fight and would lose manpower.
In the real world an outnumbered force could often win in their own homeland, for example: https://en.wikipedia.org/wiki/Swedish_War_of_Liberation
Maybe the real problem is that in EU4 it is impossible to win when outnumbered. Maybe troops have a bonus (5%? 10%?) fighting inside their borders?
The World Conquest crowd will be displeased as always as this would make conquests harder, like they were in real life...
- 1
- 1
It has been proposed several times that armies should only be able to receive reinforcements in their own countries, and in non-separatists provinces. In this way this will help to defend its own country, and also will make the current tactic of running far away to siege an enemy province totally unefficient. Of course such a change will need a complete rework of the AI. An outnumbered country should now wait in its own country, avoiding battles, maybe by hiding behind forts if possible, and wait for the invader troops to be reduced by attrition until they are weaker than him.
On the other hand, the AI offensive part should also be reworked to take this into account. Right now one of its main issue is that it takes in account the strength of the potential enemy and its allies, but not the location of the troops of both sides, so there is no preparation for war and troop positionning by the AI.
But under current rules, leaving your country when it is invaded by superior forces and invades enemy land is probably the best the AI, and even a player, can do in most cases. Sniping enemy armies was easier in real life then in EU4, where the battles last so long that the main enemy army will usually join the battle before it is won, or at least will catch your troops before you can retreat after crushing a small enemy army. It is a difficult and risky business, even for a player, and usually less profitable than carpet siege the enemy or target a weak ally of it.
On the other hand, the AI offensive part should also be reworked to take this into account. Right now one of its main issue is that it takes in account the strength of the potential enemy and its allies, but not the location of the troops of both sides, so there is no preparation for war and troop positionning by the AI.
But under current rules, leaving your country when it is invaded by superior forces and invades enemy land is probably the best the AI, and even a player, can do in most cases. Sniping enemy armies was easier in real life then in EU4, where the battles last so long that the main enemy army will usually join the battle before it is won, or at least will catch your troops before you can retreat after crushing a small enemy army. It is a difficult and risky business, even for a player, and usually less profitable than carpet siege the enemy or target a weak ally of it.
- 6
- 1
why did you ignore literally the next line where I said "if that isn't an option, then I would do X" lol