AI attacking city states every game is frustrating

[Flaxton]

Play as Australia and enjoy liberating them.

i.e. As I just did on deity and spent the vast majority of the game in +100% production for the cultural win.

 

[Ferocitus]

Play as Australia and enjoy liberating them.

i.e. As I just did on deity and spent the vast majority of the game in +100% production for the cultural win.

But that requires developing tactics and counter-strategies! It's much easier to
get Firaxis to restructure the game so that everyone can win at Deity level.

 

[sixty4half]

In every game i have played since Rise and Fall. the AI has been relentless in attacking city states. I find this to pretty game breaking and ultimately I have stopped playing for the time. Its kind of frustrating playing at all with this inevitability along with the all the other poor AI behavior. Have the developers addressed this at all? i dont really pay attention to the Civ community but this has ruined my experienced to the point i feel i have to express my concern with the small chance some change can occur.

thanks

I find it refreshing. Civ V had to many CSs at the end game. There just aren't that many Leichtensteins and Monacos left in the world (per capita). Of course CSs had a different purpose in CiV. Variety is the spice of life (William Cowper) and I'm glad that the behavior of and toward CSs is different in this game.

 

[x2Madda]

I have no issue with it. OP I understand your fustration and mods have already been recommended so outside of that you either liberate them so they are your friend for life (until they get taken over again) or enable more CS so some are left by the end.
The gameplay way to help city states is to send them a trade route. The extra gold they earn goes into military. You can also station your troops in their borders and set them to defend position forever. Sure the AI can knock the city down to 0 health but if a melee unit cannot enter the city, it cannot be taken over.

 

[Victoria]

The gameplay way to help city states is to send them a trade route.

Or take over their troops and upgrade them/fight with them. Just converting a galley to a caravel can make a huge difference

 

[TheMeInTeam]

But that requires developing tactics and counter-strategies! It's much easier to
get Firaxis to restructure the game so that everyone can win at Deity level.

? Firaxis has been doing this since Civ 5 .

I've seen suggestions to just put scouts in the way so the AI can't take it, which leaves AI in a confused long-term war scenario. Smooth implementation. 5/7 would scout again.

 

[Archon_Wing]

Since envoys scale, it's even better to just let them capture the CS and just declare war on them. Their units are weakened and you can kill them easily. If an emergency fires, that's probably game winning. Everyone else also loses their envoys as well.

It does feel like kind of an exploit. If the AI throws the game by repeatedly capturing the same CS that they can't hold; I just liberate it and wipe them out without any warmonger penalties. Stuff like scout blocking or protectorate wars seem like a waste of time to me. You only need 3 envoys (and more than anyone else) anyways aside from blue CS's.

 

[kb27787]

I would have liked some variation in how AIs approach CS... Shaka, Genghis I can understand, but when Teddy, Robert, and John do it it does ruin the flavor a bit... AIs should at least behave according to their agendas...

Sending their starting warriors all at the nearest target seems to be the default for every AI for some reason...

 

[Haig]

I'm kinda surprised that for many ppl this seems a big deal.

There's protectorate war option that you can use to step in to defend your favorite CS, plus you can reconquer them.

Also it triggers the Emergency sometimes which is cool.

I understand it bugs some people but personally I like it because it gives more life and events to the world.

 

[tedhebert]

I don't mind that the AI or the player can attack and take over CS... I too think it's an interesting aspect of the game. What most people don't like is when 10 of them are taken by the time you leave ancient era, before you
even had time to explore enough to meet them... IMO they are just too easy to conquer in the early game... A wall at game start makes it a more difficult, but not impossible proposition.

I disagree that it makes the game less complicated to give them the wall... In fact, it might make it harder, as the AI might concentrate a little more on the player instead of the CS in early game.

After the first few eras, I agree with most here that you DO have options to intervene if you want to... I just wish that 30-40% weren't already gone by the time these options are available.
Of course, it's never too late to liberate one, but...

anyways... Just installed the wall mod to try it out and see what real changes occur to my games because of it... I don't like to speculate without trying forst hand

 

[Trav'ling Canuck]

This is based on only 6 observations, so please take the small sample size into account before placing much (any) relevance on these statistics, but from the Deity level test games I've run to date, here is the average number of City States still alive at various points (based on Standard map size so 12 starting City States):

Average Turn of 1st City State conquest: T24
Avg. City States surviving at T50: 9
Avg. City States surviving at T100: 6.4
Avg. City States surviving at game end: 4

This is based on me as the human player not intervening to prevent City State conquest or to liberate them from their conqueror.

EDIT ADDITION:
And if you care about spread of observations (again, very small sample size of only 6 test games):

Earliest observed 1st City State conquest: T17
City States surviving at T50: high of 11, low of 7
City States surviving at T100: high of 9, low of 2
City States surviving at game end: high of 8, low of 1

 

[tedhebert]

@Trav'ling Canuck My god, this is even worse than I expected... Sorry everyone who keep saying this is all normal and good, but... it most definitely is NOT !

 

[orasis]

I like it. Just like barbs this adds some excitement and choice to the early game. As much as I liked V... the early game was very slow and usually boiled down to clicking next turn into the early middle ages.

Now I have waves of barbs to deal with and the threat of war... depending on how useful AI VS CS is to my long term strategy.

 

[Kruos]

This is based on only 6 observations, so please take the small sample size into account before placing much (any) relevance on these statistics, but from the Deity level test games I've run to date, here is the average number of City States still alive at various points (based on Standard map size so 12 starting City States):

Average Turn of 1st City State conquest: T24
Avg. City States surviving at T50: 9
Avg. City States surviving at T100: 6.4
Avg. City States surviving at game end: 4

This is based on me as the human player not intervening to prevent City State conquest or to liberate them from their conqueror.

EDIT ADDITION:
And if you care about spread of observations (again, very small sample size of only 6 test games):

Earliest observed 1st City State conquest: T17
City States surviving at T50: high of 11, low of 7
City States surviving at T100: high of 9, low of 2
City States surviving at game end: high of 8, low of 1

From my observations, I confirm these numbers. At Deity level the CS fall like flies. On average, only a handfull of them are still alive at late game (around T150) and until the end. For me, as I am used to it now, it's not that much a problem, but I fully understand players who are disturbed/annoyed by this AI behavior, espescially as it harm a game design concept and the gameplay around it.

This AI behavior is the result of the improvment of the AI efficiency, and it is a good thing at the end. So IMO, it is out of question to change this. A possible 'fix' IMO would be to 1) make CS more difficult to conquer and 2) add some 'protection lever' for the CS suzerain.

For the 1), the dev already tried some little things, but now we can clearly say that it was not enough. I know that there is a mod that add wall to capitals from the start, maybe it's too much radical, maybe not. But I think it is a good direction to follow. Maybe just give a defensive bonus to palace building, or some bonus HP. Definetely need more of something.

For the 2), I would propose to make a suzerain go to war if one of its CS is attacked. Is it not the purpose of the 'suzerain' role, to protect its vassals? It would make so much sense! With such mechanism, one would think twice before attacking a CS with a suzerain, and obviously some numbers would have to be added in the AI behavior tree for the 'go to war with a protected CS' decision.

Again, I repeat, for me the current state is OK, even if I deplore it because it harm a game design concept and the gameplay around it.

 

[schondette]

I like it. Just like barbs this adds some excitement and choice to the early game. As much as I liked V... the early game was very slow and usually boiled down to clicking next turn into the early middle ages.

Now I have waves of barbs to deal with and the threat of war... depending on how useful AI VS CS is to my long term strategy.

how is it exciting watching AI attack city states? i can understand if you could do something about it but you cant. They are on the map to be able to compete for resources over, not for free city plots which is exacerbated by other AI behavior issues with the game. eg city spamming in nonsensical locations.

 

[tedhebert]

From my observations, I confirm these numbers. At Deity level the CS fall like flies. On average, only a handfull of them are still alive at late game (around T150) and until the end. For me, as I am used to it now, it's not that much a problem, but I fully understand players who are disturbed/annoyed by this AI behavior, espescially as it harm a game design concept and the gameplay around it.

This AI behavior is the result of the improvment of the AI efficiency, and it is a good thing at the end. So IMO, it is out of question to change this. A possible 'fix' IMO would be to 1) make CS more difficult to conquer and 2) add some 'protection lever' for the CS suzerain.

For the 1), the dev already tried some little things, but now we can clearly say that it was not enough. I know that there is a mod that add wall to capitals from the start, maybe it's too much radical, maybe not. But I think it is a good direction to follow. Maybe just give a defensive bonus to palace building, or some bonus HP. Definetely need more of something.

For the 2), I would propose to make a suzerain go to war if one of its CS is attacked. Is it not the purpose of the 'suzerain' role, to protect its vassals? It would make so much sense! With such mechanism, one would think twice before attacking a CS with a suzerain, and obviously some numbers would have to be added in the AI behavior tree for the 'go to war with a protected CS' decision.

Again, I repeat, for me the current state is OK, even if I deplore it because it harm a game design concept and the gameplay around it.

I strongly disagree... now, this is on what I would call an 'irrelevant' sample size, but I tried the walls to CS mod this weekend, and I got trashed 3 times at immortal by turns 30/40... granted it was 3 times with the same game which was a very bad map start with 3 civs very close to me, but even knowing it was coming by replaying the same start, I still got trashed by Mapuche attacking me with 6 warriors combined with Spain attacking from the other side with 3 warrios and 2 slingers...

So ok, yes it was a very bad start, and yes I'm not the best player around either, but what this tells me is that games would be VERY much harder on higher levels if the AI concentrated on the player instead of concentrating on CSses... isn't this what everyones been complaining about, the AI not being a challenge ?

 

[Kruos]

I strongly disagree... now, this is on what I would call an 'irrelevant' sample size, but I tried the walls to CS mod this weekend, and I got trashed 3 times at immortal by turns 30/40... granted it was 3 times with the same game which was a very bad map start with 3 civs very close to me, but even knowing it was coming by replaying the same start, I still got trashed by Mapuche attacking me with 6 warriors combined with Spain attacking from the other side with 3 warrios and 2 slingers...

So ok, yes it was a very bad start, and yes I'm not the best player around either, but what this tells me is that games would be VERY much harder on higher levels if the AI concentrated on the player instead of concentrating on CSses... isn't this what everyones been complaining about, the AI not being a challenge ?

Could you please precise with what you 'strongly disagree', reading your post it's not clear to me. You seems to say that AIs rushing players would be good to improve the difficulty, which is a fair opinion, but not very related to my post's points (nor the topic thematic I would say).

The 'wall to CS' mod is maybe a little bit to radical as I said in my post, it is why I propose instead a defensive buff (strengh, hp, ..) linked to the palace building present in all capitals (players and CS). It would be a smoother solution I think.

 

[tedhebert]

@Kruos Sorry... Strongly disagree with your statement that's it's 'out of the question to change this' ;-)

To add more detail, I would rather see the added AI efficiency used against the player than used against the CSes

 

[Trav'ling Canuck]

I haven't carefully documented the circumstances of City States that get conquered versus those that don't, so this is just guesswork on my part from what I have observed, but the three factors (other than human player intervention) that appear to most strongly affect whether a City State survives or not are:

1. Geography: Starting with a Mountain range protecting one side of their city or starting on an island appears to significantly increase their longevity. Partially because they're located later, partially because they're harder to successfully attack, and partially as an overlap with item 2 below.
2. Number of neighbours: City States with a single AI neighbour seem to be left alone more than City States are between two or more AI Civs. I don't think this is just a 50% reduction based on the number of AI civs who could possibly attack them. I think (but cannot prove) that there may be a tendency for the AI to be more likely to attack a City State that moves them closer to another AI's civ's border.
3. Number of Inter-AI Wars: When they capture a City State from a neighbouring AI during a war, the AI appears to prefer to liberate the City State rather than keep it (even if they subsequently attack the City State later). Thus, warfare between the major AI civs sometimes results in some of the initially captured City States being subsequently liberated. The slowdown in the pace of City States disappearing after T100 in the test games I've observed appears to have more to do with the liberation of some of the previously conquered City States than with any slowdown in the number of City States being attacked and conquered.

 

[UWHabs]

I haven't carefully documented the circumstances of City States that get conquered versus those that don't, so this is just guesswork on my part from what I have observed, but the three factors (other than human player intervention) that appear to most strongly affect whether a City State survives or not are:

1. Geography: Starting with a Mountain range protecting one side of their city or starting on an island appears to significantly increase their longevity. Partially because they're located later, partially because they're harder to successfully attack, and partially as an overlap with item 2 below.
2. Number of neighbours: City States with a single AI neighbour seem to be left alone more than City States are between two or AI Civs. I don't think this is just a 50% reduction based on the number of AIs who could possibly attack them. I think (but cannot prove) that there may be a tendency for the AI to be more likely to attack a City State that moves them closer to another AI's border.
3. Number of Inter-AI Wars: When they capture a City State from a neighbouring AI during a war, the AI appears to prefer to liberate the City State rather than keep it (even if they subsequently attack the City State later). Thus, warfare between the major AI civs sometimes results in some of the initially captured City States being subsequently liberated. The slowdown in the pace of City States disappearing after T100 in the test games I've observed appears to have more to do with the liberation of some of the previously conquered City States than with any slowdown in the number of City States being attacked and conquered.

Geography definitely makes a difference. If the City-state is protected (ie. lake, or needs a mountain pass to get to), then they'll probably survive. Also, if they're far enough away from any AI, then they have time to build units/walls to defend themselves.
I don't think the number of neighbours matters too much. I think it depends more on whether any of the close neighbours decides to take out city-states instead of other players. For example, in my current game, it was me+Scotland on an island together with 4 city-states. For once, Scotland decided not to attack me, but they did capture 2 city-states.
The third point is a weird one. I remember a game maybe last fall where the AI captured a city-state, and then I declared war on them and captured it and decided to keep it. And then they declared war on me again later, and ended up capturing back the city-state, but this time liberating it. So they're coded to capture city-states, but they're also coded to liberate them.

Maybe the best solution is to live with the AI wiping out ~half the city-states in the game at higher levels, but you should start with more. So on a small map, instead of there being 6 civs and 9 city-states, maybe we should just have 6 civs and 12 city-states. That way you can still lose half and have enough left to play with. If they also adjusted the map scripts to try to always place 1-2 of them on like off-shore islands or island groups, that could also help.