AI's Inability to Defend Its Territory Is Painful

[Chlodio]

As Parni I noticed Seleucids and their vassals were fighting somewhere in the east, their opponent wasn't Maurya, but some insignificant foe. Seleucids themselves had 40 cohorts, a shit ton of manpower, and their subjects could together bring in more than 100K troops.
Meanwhile, all had was 15 cohorts, so I decide to gamble. It quickly becomes apparent everybody was in the east, and I occupied all of Parthia, part of Bactria, parts of Iran (including Seleucid capital) virtually unopposed. All they did was occasionally send armies 1–3K to skirmish... After nearly a decade of war, they finally began bringing bigger armies, but at that point, I had so much war score and their war enthusiasm was so low I could peace out without having to fight superior numbers.

This all happened because they:

1. insists on taking all their forces to war and leaving their territory defenseless
2. insists on finishing their war before defending their country, no matter what occurs in their own turf meanwhile

I realize AI programming is hard, but this just isn't good. If not anything else, AI should very least recall all their troops the moment their capital falls.

 

[Herennius]

This problem is also fueled by the tendency of big empires not to built or to dismantle existing fortifications in most parts of their country. Forts alone - unless very high level - are unlikely to stop a invasion by themselves, but they could buy the AI needed time here.

 

[Cadzon]

I get your point. But I've played Seleucid as well and it's hard to have a 2 front war with so large distances. I was against maurya and phrygia and I didn't have enough on either side if I split them up. I can only imagine it's even harder for the ai to decide what to do. In your case they probably had an easier choice. In my games they usually respond with minor delay when I do your strategy, and VERY often they peace out the other opponent earlier than expected, and go all in on me... And I've played almost 1000 hours I just noticed... Jesus

 

[CaptinObvious]

This problem is also fueled by the tendency of big empires not to built or to dismantle existing fortifications in most parts of their country. Forts alone - unless very high level - are unlikely to stop a invasion by themselves, but they could buy the AI needed time here.

Check this out:

I checked AI fort stuff...and....take a look:

NAI = {
       [...]
       #economy
       FORT_BUDGET = 10
       [...]
       }

what this means is that the AI never spends more than 10% of GDP on forts, doesn't sound too bad...right?

well, check this monstrosity out:

NAI = {
     [...]
    #Forts
    PLACE_FORT_BORDER_FACTOR = 1.5 #Is on border 1/0 multiplied by this.
    PLACE_FORT_COAST_FACTOR = -1.5 #Is on cost 1/0 multiplied by this.
    PLACE_FORT_PORT_FACTOR = 2.0 #Is on cost 1/0 multiplied by this.
    PLACE_FORT_DISTANCE_FACTOR = 1 #Distance in provinces to nearest fort multiplied by this.
    PLACE_FORT_CAPITAL_FACTOR = 1 #Capital 1/0 multiplied by this.
    PLACE_FORT_TERRAIN_FACTOR = 1 #Def bonus multiplied by this.
    PLACE_FORT_RIVER_FACTOR = 1 #River 1/0 multiplied by this.
    PLACE_FORT_LEVEL_FACTOR = 0.5 #Existing fort level multiplied by this.
    PLACE_FORT_PROVINCE_CAPITAL_FACTOR = 3 #Forts in province capitals.
    PLACE_ROAD_MIN_DISTANCE = 2 #Minimum distance from one road endpoint to another.
    DISBAND_FORTS_CAPITAL_BIAS = 2 # More likely to choose Forts in capital city to disband first
    [...]
    }

what does this mean?

It means that the AI not only doesn't build enough forts, but has been *intentionally* instructed to *demolish* capital forts, ever seen the AI just leave it's capital undefended? well, this is why...

this is what happens when we have a design-wise disaster, the AI isn't dumb, it' just been designed poorly

 

[Akbar The Great]

AI's painful full stop.

PDS answer to all its games problem is, "but more content" instead of spending time releasing games that are actually ready and have actually been thought-through and thoroughly tested, they ignore what's broken, and just pile more poorly realised updates/DLC/patches on top.

They've had the last release day nickel they'll ever see from me after the IR disaster (and proven again with the disaster that is Emporer). think people are finally beginning to have enough of PDS's model.

1. Release a half ass product.

2. Fix it, sort of, well a bit - ignore what's too difficult and move on.

3. Pile on more DLC and hope everyone forgets what's already broken.

Repeat.

 

[Efimer]

... the AI not only doesn't build enough forts, but has been *intentionally* instructed to *demolish* capital forts...

The next game I'm going to fill all the building slots of my capital with forts...

 

[Herennius]

The next game I'm going to fill all the building slots of my capital with forts...

Sounds like a plan - just make sure that the capital doesn't grow to big or you will eventually just run out of time

But seriously, I'm currently going at least a bit in that direction - I'm playing Trapezous and sharing a border with Phrygia, I think a lvl 8 fort in the capital can't hurt:

I wish that the AI would fortify important spots in a similar way to have more of a challenge. The Gold (in which most AI empires drown anyway) would be wisely invested.

 

[Jays298]

As Parni I noticed Seleucids and their vassals were fighting somewhere in the east, their opponent wasn't Maurya, but some insignificant foe. Seleucids themselves had 40 cohorts, a shit ton of manpower, and their subjects could together bring in more than 100K troops.
Meanwhile, all had was 15 cohorts, so I decide to gamble. It quickly becomes apparent everybody was in the east, and I occupied all of Parthia, part of Bactria, parts of Iran (including Seleucid capital) virtually unopposed. All they did was occasionally send armies 1–3K to skirmish... After nearly a decade of war, they finally began bringing bigger armies, but at that point, I had so much war score and their war enthusiasm was so low I could peace out without having to fight superior numbers.

This all happened because they:

1. insists on taking all their forces to war and leaving their territory defenseless
2. insists on finishing their war before defending their country, no matter what occurs in their own turf meanwhile

I realize AI programming is hard, but this just isn't good. If not anything else, AI should very least recall all their troops the moment their capital falls.

I can imagine it's hard to program an AI to balance between defending a vast territory and fielding an offensive army.

In my game as Rome, same thing happened with Carthage multiple times. They'd be steamrolling some insignificant tribes in southern Iberia. But for some reason the war goes on for years.

And Massilya (their neighbors in Africa) were steam rolling Carthage in Africa each time.

The AI and even myself have gotten used to using vassals as garrisons!

And when I did join the war, there was no army around Carthago! Only their vassal storm nearby.

I think the AI is trained to fight offensively. And that works well for border wars. But not for overseas powers.

It almost felt like cheating because I was able to sack and annex Carthago. And to top it off their Navy had no morale.

And despite taking a ton of land very little is annexed by the AI.

 

[Chlodio]

I can imagine it's hard to program an AI to balance between defending a vast territory and fielding an offensive army.

I recall reading EU4 changelog where they said they told AI specifically to reserves a certain percentage of their troops for the defense.

 

[GAGA Extrem]

Sounds like a plan - just make sure that the capital doesn't grow to big or you will eventually just run out of time

But seriously, I'm currently going at least a bit in that direction - I'm playing Trapezous and sharing a border with Phrygia, I think a lvl 8 fort in the capital can't hurt: [...]

*coughs*

(To be fair - it's probably a bit harder these days. It's also the reason why it is harder. )

 

[Chlodio]

*coughs*
View attachment 594424
(To be fair - it's probably a bit harder these days. It's also the reason why it is harder. )

I don't understand why you don't just multiply the cost of each level, so that level 1 fort would cost 50 gold, level 2: 100, level 3: 150, so on and so forth.

 

[Mike6979]

Also, Paradox's idea of difficult war AI is carpet siege. I was frustrated by this at first but then eventually I built 8 horse archers and 2 LC (flavour, sue me.) and I can split that into smaller stacks as needed, it never leaves my land and kills the carpet siege nicely. At that point all the carpet siege is doing is killing the AI army, and giving me carpal tunnel.

It is amazing in history how you do not hear about carpet siege.

 

[Dayvit78]

You can also chalk it up to all wars becoming total wars - this is true across all Paradox games, where only HOI4 and Vicky 2 should have that aspect.

There needs to be some differentiation between troops at home (and garrison forces) vs offensive forces. With the latter being exponentially more expensive to maintain (gold or attrition or whatever). In EUIV, you regularly see the European powers sending stacks of 100,000 to the East Indies. And the same goes for here. The Seleucids could raise massive armies and project them into neighboring territories, agreed; but that didn't mean the rest of their country had 0 troops left.

If the AI can prioritize forts where, they could also prioritize armies. However, the player would probably be able to exploit it because the player also has a total war mentality.

 

[jurebaao]

As Parni I noticed Seleucids and their vassals were fighting somewhere in the east, their opponent wasn't Maurya, but some insignificant foe. Seleucids themselves had 40 cohorts, a shit ton of manpower, and their subjects could together bring in more than 100K troops.
Meanwhile, all had was 15 cohorts, so I decide to gamble. It quickly becomes apparent everybody was in the east, and I occupied all of Parthia, part of Bactria, parts of Iran (including Seleucid capital) virtually unopposed. All they did was occasionally send armies 1–3K to skirmish... After nearly a decade of war, they finally began bringing bigger armies, but at that point, I had so much war score and their war enthusiasm was so low I could peace out without having to fight superior numbers.

This all happened because they:

1. insists on taking all their forces to war and leaving their territory defenseless
2. insists on finishing their war before defending their country, no matter what occurs in their own turf meanwhile

I realize AI programming is hard, but this just isn't good. If not anything else, AI should very least recall all their troops the moment their capital falls.

I know what you mean. they can't attack either.

Once i was fighting east and south and got invaded by north. While my tributary stayed iddle i was invaded by 2-3 stacks of 20k army, while i had barely 25k with my armies.luckily i had a 12 martial general and i managed to beat a 18k stack in the way and 2-3 8-12k armies. then i went to the settlement of fuccens, wich back there was part of one of my vassals. and wha did happen?

1- my enemies started attacking me like animals, they did not bother regrouping and consolidating its armies, they just threw themselves across and narrow path across a mountain, and with that all desavantage they attacked me with 60k troops, a little stack of a time. killed tens os thousands, lost so few.
Ai is really really reallly dumb

 

[Herennius]

If you were Armenia, currently in a war vs. Persia, Trapezous and some other states...a war you are badly losing...would you tear that fort down?

No? Well, the AI thinks different about it as it seems. I took the freedom to file a bug report, though I'm aware that it is probably not a bug in a strictly technical sense:

AI tears down border fort right at the front during a war

1.4.2 I'm at war with Armenia and suddenly one of their forts at my border vanishes... Since I don't know of any random event destroying a fort, I suspect the AI has torn it down - maybe because of their rigid fort budgets....but I think we...

forum.paradoxplaza.com

 

[Talib123]

I think a good way of reducing the issue would be to introduce city walls as a building.

Sometime I even think that it could also be possible to "hard set" some forts. Just letting players and AI decide wether they want to maintain it or not.
I think of choke points, historical important or verified forts like megara, korinth, athens, rhodes, carthage and so on and so on.

 

[Ketchup & friends]

This is closely related to another problem with the AI: it treats any war as an all-or-nothing total war and sends their entire army into what should realistically be minor border conflict over a couple of towns.

To solve both problems, AI priorities need to be adjusted.

here are our ideas on such adjustments

I. Provincial troops:

Starting with a certain size, AI should assign a significant portion of the troops to Regional Governors as regional armies.

AI should ensure that EVERY non-capital Region that borders

• a foreign power
• a navigible tile
• a Barbarian Hotspot

has an army assigned to the governor that is able to deal with a small invasion force.

Together with Forts this this should serve as a first defensive line.

That would be

• Realistic, that’s how large ancient empires like Rome defended their borders IRL.
• Balanced, large AI would tie up lot of troops as Regional Garrisons.
• Provide some intrigue, as it increases the powerbase of Governors.

If it is overrun, then, the AI should send the main army to deal with this problem.


II. Cut losses.

If an AI-controlled country is fighting multiple wars AND losing two or more of them, it should offer peace out of one of them. Preferably, trying to get out of aggressive wars it started before surrendering to those who attacked them.

If the enemy accepts, it should try to turn the table in the other front. If after a while it’s still losing two or more wars, repeat that step.

If the enemy refuses, AI should try the same in one of the other wars.


These two ideas would ensure the AI acts somewhat reasonably.

 

[Dan the cynical barbarian]

This is closely related to another problem with the AI: it treats any war as an all-or-nothing total war and sends their entire army into what should realistically be minor border conflict over a couple of towns.

To solve both problems, AI priorities need to be adjusted.

here are our ideas on such adjustments

I. Provincial troops:

Starting with a certain size, AI should assign a significant portion of the troops to Regional Governors as regional armies.

AI should ensure that EVERY non-capital Region that borders

• a foreign power
• a navigible tile
• a Barbarian Hotspot

has an army assigned to the governor that is able to deal with a small invasion force.

Together with Forts this this should serve as a first defensive line.

That would be

• Realistic, that’s how large ancient empires like Rome defended their borders IRL.
• Balanced, large AI would tie up lot of troops as Regional Garrisons.
• Provide some intrigue, as it increases the powerbase of Governors.

If it is overrun, then, the AI should send the main army to deal with this problem.


II. Cut losses.

If an AI-controlled country is fighting multiple wars AND losing two or more of them, it should offer peace out of one of them. Preferably, trying to get out of aggressive wars it started before surrendering to those who attacked them.

If the enemy accepts, it should try to turn the table in the other front. If after a while it’s still losing two or more wars, repeat that step.

If the enemy refuses, AI should try the same in one of the other wars.


These two ideas would ensure the AI acts somewhat reasonably.

This are some really good ideas, however I do fear this possible governor mechanic becoming the so hated Stellaris sectors as depending on the AI is problematic

 

[Chlodio]

I think a good way of reducing the issue would be to introduce city walls as a building.

Sometime I even think that it could also be possible to "hard set" some forts. Just letting players and AI decide wether they want to maintain it or not.
I think of choke points, historical important or verified forts like megara, korinth, athens, rhodes, carthage and so on and so on.

After reading more about conquest during this period, I'd think it would more realistic if:

• fort buildings would only be constructible on the border and have a garrison of a hundred men
• all provincial capitals would automatically have a garrison of a thousand men, and nation capitals five thousand
• the fall of provincial capital would auto-occupy all the provinces within the province
• you could only advance to hostile adjacent provinces if you have captured the provincial capital

 

[KEV1NJ]

This!!!

After reading more about conquest during this period, I'd think it would more realistic if:

• fort buildings would only be constructible on the border and have a garrison of a hundred men
• all provincial capitals would automatically have a garrison of a thousand men, and nation capitals five thousand
• the fall of provincial capital would auto-occupy all the provinces within the province
• you could only advance to hostile adjacent provinces if you have captured the provincial capital

Also you are all ignoring the elephant in the room. Paradox cares only for the MP players, even though (I suspect) the solo players make up as much or more of their player base. They have repeatedly unbalanced the SP aspects of their games in favor of trying to balance MP. I doubt there is any interest there in fixing the (alleged) AI.