AI rarely captures cities?

[Gidoza]

It's possible that it's simply too easy to defend under the current balance.

I find that defending gets considerably easier the later the game goes on. In large part this is due to the very high military cap that appears later on in the game - as Tradition I often have little to no trouble at all defending even a fairly large area with a meagre number of units against the onslaught of a warmonger with upteen cities - ultimately, the warmonger may have tons more military, but is unable to use them against me in any meaningful way because he can only fit 1 unit per tile. Earlier in the game when it's more of a hassle to train units in general, it's easy to be surrounded when fighting a meagre opponent's army when unprepared.

I'd figure that if the unit cap were reduced proportionally all around (e.g. warmongers have fewer units but can actually use them all; a tall civ's cap is so low that it is difficult to get enough units to actually patrol the border), I think we'd see much more interesting warfare. This is besides the point that fighting walls of units is boring and makes turns take forever...so that just adds on to the "late game" discussions that some people have had and why we get games of the first 150 turns so much more often - because the first 150 turns of warfare is just more fun but also more balanced.

 

[CppMaster]

I'd figure that if the unit cap were reduced proportionally all around (e.g. warmongers have fewer units but can actually use them all; a tall civ's cap is so low that it is difficult to get enough units to actually patrol the border), I think we'd see much more interesting warfare. This is besides the point that fighting walls of units is boring and makes turns take forever...so that just adds on to the "late game" discussions that some people have had and why we get games of the first 150 turns so much more often - because the first 150 turns of warfare is just more fun but also more balanced.

Hmm I like this idea. Make the cap lower, but at the same time make units cost more and have more strength, so each unit matters more. Or alternatively, just make city defenses weaker, it would achieve similar things, but easier.

 

[looorg]

I don't doubt that one of the issues that the AI is so bad at war is that they do have to many units. It becomes nearly impossible for them to move in an effective manner. It also makes the players war vs them weird as it mostly is the same every time -- you defend and range them to death and when they start to run out of zombie horde moves you move in and take what you like repeat. It's like being at war with a rival and also together with another ai. Trying to move to friendly-ai land to reach enemy-ai land is a pain as there is literally a unit in more or less every tile just shuffling about.

That said I don't really have a good solution for it. Even if one could implement some kind of CIV6-army-corpse deal that would just mean they would have more of them and not less units probably. Making cities weaker is not really a good idea either, the cities are already kind of weak and the AI-s "new" tactic since a few versions back appear to be to encircle their own cities and then just sit there and wait to be bombarded to death from range.

The only area where AI war appears to actually be working is naval war and attacks vs coastal cities. Here the AI seems to be working and doing fairly well, certainly vs other AI at least. But land war is just a gigantic-zombie-horde-mess of units that make every war look like trench warfare ala WWI. Going on forever but rarely moving an inch forward.

 

[CppMaster]

Making cities weaker is not really a good idea either, the cities are already kind of weak

Doesn't this topic refute that? Maybe the cities are not weak at all. Compared to Civ6, I think cities in VP are very strong.

 

[Rekk]

but at the same time make units cost more and have more strength, so each unit matters more.

What is your intention when you say "more strength"? All units deal/take less damage so that they stick around for longer?

 

[looorg]

Doesn't this topic refute that? Maybe the cities are not weak at all. Compared to Civ6, I think cities in VP are very strong.

Yes and No I would say. The reasons the AI is so bad at taking cities from each other is the horde of units. I posted some screenshots on the first page of an AI vs AI battle and there is just not a spot to stand in (my para has to stand on a mountain just to fit in there and get vision). So they just never really get to the cities. If the AI did what the human player do with setting up some proper bombardment and ranged units then the cities are actually quite weak. If you get your cannons in range (or range3) the cities don't really last that long. It's sort of what happens to coastal cities where the AI can actually move units in and out more efficiently then they do on land. They bombard and melee that fine. They just go all zombie-horde with the land units and move units willy-nilly. I mean who the heck puts ranged and siege units in the front line etc. So the cities are strong cause land-AI is bad at troop management and deployment. Not cause cities are actually that strong.

 

[Fluffball]

Again working off my personal experience i feel having the AI build roads/railroads on its conflict borders would help solve a lot of the unit spam log jam for the AI and seemingly moving units in a sub-optimal way, especially in movement restricting terrain. With roads and especially railroads which come in to play when there are very large numbers of units i can station my back up units far from the front to the point that for fast moving units in particular i often move them to cities in the far reaches of my empire to heal so they heal more quickly from being in a city.

Without roads/railroads i have to very carefully place my units so i can safely replace damaged units and this often means i have to station melee units in positions where ranged units could go to ensure i can reach the spot i may need to swap with and even then it often involves a bit of careful shuffling often needing to move 2-4 additional units to move the 2 i actually want to swap like some sliding puzzle game. Once i have roads and especially railroads i have lots of options available to swap units easily and can have more ranged units able to fire in support.

I don't know how many moves ahead the AI thinks but i assume it isn't very far as that dramatically increases processing so it may move units poorly as it can't think ahead far enough to move the unit where it actually wants to and especially so when it involves moving another unit to do so as is seen when you give a unit orders to move a number of turns away and it keeps waking up because it's next position is blocked or if you watch them carefully (with the movement display mod being useful for this) they will often try to solve the block by constantly moving back and forth on the same spot.

 

[CppMaster]

What is your intention when you say "more strength"? All units deal/take less damage so that they stick around for longer?

I meant more Combat Strength compared to cities CS. Units vs Units could stay the same, but Units vs Cities would be better.

So it's basically like corpuses in Civ6. They have more CS, but they also cost more. It'd make more concentration of power per tile and moving armies would be less cumbersome, because there would be less units overall, because of cost, limit and less need for many units.

However, like I've mentioned before, just nerfing cities CS and lovering supply cap would result similarly, while having much less things to balance.

 

[Rekk]

I meant more Combat Strength compared to cities CS. Units vs Units could stay the same, but Units vs Cities would be better.

So it's basically like corpuses in Civ6. They have more CS, but they also cost more. It'd make more concentration of power per tile and moving armies would be less cumbersome, because there would be less units overall, because of cost, limit and less need for many units.

However, like I've mentioned before, just nerfing cities CS and lovering supply cap would result similarly, while having much less things to balance.

The main thing is that if all you're doing is reducing city CS and increase unit cost would kill each other as quickly as they would now. The AI would very quickly run out of units (like it currently does at low difficulty levels), and have a hard time replacing them due to the increased production cost.

 

[CppMaster]

More concentration of power per tile means that more total CS would actually strike the city per turn and less units would clutter doing nothing in the back. Also there would be less units to manage overall. If the defender killed many units, that means that he also had many units, because cities are weaker, so that's fine. Simply put, the attacker would need a smaller army advantage in order to capture a city compare to what we have now, which would result in more often cities captures.

 

[andersw]

I agree cities have too much hp/defence.
But Im not sure how much less they should have and if its mostly HP or defence that is the main issue.
Edit: I would say this is mainly from castle and onwards.

 

[Rick Drayson]

I too began seriously noticing this issue
All because i began a series of games with max amount of AI packed in. No AI was able to get ahead because none could conquer cities
This ones again comes back to the insane decision to prevent unit stacking.
They went from too many units on the same tile in civ IV and jumped straight to just one. Bad decision and very foolish with lots of ramifications that really get noticed when the AI got beefed and unit numbers exploded.

I see that they tried to address the issue in Civ VI.
My biggest bugbear was not giving the players a choice. Limit unit stacking to 1->infinity.

ps: ok the main issue with ai not taking cities is the ai not knowing how to use bombardment units. That is what they are there for and the AI doesn't use them.

 

[SullySaltlicker]

From my experience, a contributing factor to cities having so much strength, at least early game when armies are small, is that having a garrison in a city gives it 2 health bars, both of which you have to reduce to 0, and both of which can heal a bit every turn. The difference in damage between attacking a garrisoned city and one without is massive enough. But if the garrison's health is getting too low for comfort, the defending civ can just swap in a healthy unit with practically no drawback, or, if the attacker kills the garrison unit but not the city in the same turn, just buy another unit in the city for a fresh health bar.

Oh what's that, my city has lost its garrison and is ready to fall? Here let me just slap another health bar on there for you to chew through. Cry about it, nerd.

Add in Tradition buffing the CS of the city itself, and the fact that attackers have to stop attacking to heal while the city can heal and attack along with the garrisoned unit, and it feels like a Tradition civ with appropriate walls/castle/etc can use just their garrison to defend a city in some situations. I'm not saying that Tradition should lose its city buffs, but that the garrison itself is the most unbalanced part, and the Tradition buff for city CS amplifies its issues.

 

[balparmak]

I've been playing with reduced supply cap (and proportionally reduced city defense) for so long I can't even compare it to default VP, but I don't think less units help in this regard. Weakening cities can help, but I feel the main problem is AI being too cautious and not mounting large-scale offenses. It feels like AI tries to conquer/push with as few units as possible like a human would, keeping many units in reserve and not exploiting their huge advantage in numbers. This can be acceptable for non-domination AI, but domination-focused civs should be much more aggressive. I found some old posts of ilteroi about some values that may help, attaching them if anyone want to experiment. Only increased offense score bias from 20 to 50 here, I don't know that part of the code so don't know if this is enough. People wanting weaker cities can try this and provide feedback

 

[notaspambot]

I've been sort of avoiding this topic because of the difficulty in putting my thoughts into words, but I personally suggest three solutions, in order of importance:

1) Make the AI much more willing to attack cities with melee units.
2) Make war weariness directly affect production of military units.
3) Reduce the durability of cities in general, whether by strength/HP.

As I've mentioned before, I've personally taken to adding a +75% city attack promotion on AI units in my games, which has helped to some degree of success, especially with how the AI utilises melee units.

As for the supply cap, I don't agree that it's too high. You'll generally have plenty if you're seeking out to fill a single front. Make that front a bit longer though, and it's easy to start feeling a bit stretched, and you can easily outright run out of supply if you want to maintain a navy on top of that, or heavens forbid, a second front. Not to mention from the war weariness penalties that you'll have to bear through if you want to actually accomplish anything.

There can be good arguments made for messing with the supply system, especially with regards to the amount of clicking/micro involved in the later parts of the game, but not for this specific discussion.

Besides, the real issue with the AI unit swarms is not the supply cap, AKA the amount of standing units an AI has at the time of a war declaration, as much as its rate of reinforcements. It's this issue we have to tackle if we want to see more cities trading hands, whether by affecting it directly, or by making the AI better adjust to it.

 

[Recursive]

@ilteroi discovered some issues with the AI that may be blocking AI city capture attempts from occurring properly. There should be some fixes for the next version - let's see if that's the culprit first before rebalancing the game.

 

[SullySaltlicker]

@ilteroi discovered some issues with the AI that may be blocking AI city capture attempts from occurring properly. There should be some fixes for the next version - let's see if that's the culprit first before rebalancing the game.

Very nice. Will the new skirmisher concept be in the next patch as well?

 

[Borya]

I can see on GitHub that code has been added for units created in damaged cities to also be damaged.
Does this affect both built and drafted units? (I'd vote only for reducing health of drafted units)
What's the health reduction? (I'd vote for about 50% of missing city health, so that recruited units do not start with less than half amount).

 

[Recursive]

I can see on GitHub that code has been added for units created in damaged cities to also be damaged.
Does this affect both built and drafted units? (I'd vote only for reducing health of drafted units)
What's the health reduction? (I'd vote for about 50% of missing city health, so that recruited units do not start with less than half amount).

All units. It's a % reduction in HP equal to the % of max HP the city has taken in damage.

 

[Recursive]

Very nice. Will the new skirmisher concept be in the next patch as well?

Probably not.