Are we looking at CiV's AI the wrong way?

[testdummy653]

I been reading a lot of threads about AI's and I want to address them.
No, I'm not an expert programmer(or AI genius) or anything so this just an uneducated theory. on how i think it runs
One thing everyone seem to ask is.....

Why does the AI not go for other victory conditions than conquest and why don't they win at all?
Or why does the AI declare war on me all the time?


Let's think about these questions. . . . . . . (I wrote this post in a hour, so i did put some thought) Okay, I'm done thinking.

What does the AI need to do to win Culturally? Build a few cities, get a lot of culture.
What does the AI need to do to win Diplomatically? Build Cities, bribe city states, build the UN.
What does the AI need to do to win Space Victory? Research, Research, and build cities, and build Space Ship parts.
What does the AI need to do to win a Conquest Victory? Build a military, and Cities, and Conquer.

Now that all very nice and good that I wrote those brief conditions out, but think about it.

Let say that CiV gives every civ a victory to go for in the game. Imagine this 6 player map.

I will list the CiV and what they are going for victory wise.

On Continent A:
India (Culture)
France (Diplo)
Aztecs (Military)
Siam (Military)

On Continent B:
Persia (Space)
You ()

Who will win/dominate on Continent A? Well it gonna be either Aztecs or Siam. Why?
France nor India will build a strong enough military to beat Siam or Aztecs, because their victory conditions are not dependent on a military campaign (of course, you can argue survival is dependent on a military, so any other victory is dependent on military). While the Aztecs and Siam will be producing much more military units, and focusing on defeating it enemies and taking their capitals.

What if India survives the war? Can they still win?
Think of it as this: India gets DoW, and stops producing happiness building and temples, and start rush buying units, an switches production from culture buildings to units. The turns that India is at war, are turns of production of units and not focused on culture. This means that x amount turns at war, makes India wait x more amount turns for cultural victory. So yes they can win, but it become more unlikely, the longer they are at war.

Why does Persia declare war on me if it's going for a space race?
Think of it as this, the AI see you as a threat not matter what you do. Why? Because your trying to win just as much as Persia is trying to launch a space shuttle. If Persia see that your weak enough to knock out of the game or slow you down enough, they will try. But Persia will be less likely to declare to war on you though because of their goal of Space.


The AI has some problems....
The AI is so focused on its victory condition that it usually ignores any attempts of Diplomatic or Culture or Space race by the human opponent. This makes it easier for us to win.
Also the AI does not switch victory goals, in game, this is why Siam could be at Future tech, and still hasn't built the spaceship. (Siam is going for a Military Victory).
The AI has a problem shipping units overseas.

My Summary/My theory is that AI's have their own victory conditions they try for, but that the military civ are usually the most effective, and that these military civs will effect the other civs chances of achieve their goals. The more military Civs or run away civs that are in the game, the less likely you will see a victory other than conquest.

I'm actually kinda of nervous, I can see someone posting your wrong in big red letters. And then posting a huge computer script explaining whats going on, and how wrong i am. (and how much time i wasted).

If you have any question please post, i can ramble more nonsense....

I should point out that it did not occur to me that time(score) victory could also be a valid victory condition for the AI.

Edit: I'm recently doing a test case on this idea see below:
http://forums.civfanatics.com/showthread.php?p=9893699#post9893699

Also the new patch info, seem to be fixing some things, so this theory may become invalid..

 

[Baleur]

Random Personalities.
But yeah.. I just wish the AI's did NOT have omniscient future planning minds, didnt view the game as a GAME, but rather did its part in suspending our disbelief (which is the developers part of the deal, our part is to immerse ourselves into the world and pretend its real. Sid Meier himself spoke about this for an hour at the GDC keynote, yet it seems the civ5 dev team completely disregarded it all).
Instead of, predicting gunpowder, every optimal path to a specific victory, and declaring war on everyone around in order to make the next 1000 years fit their conquest objective.

I wish they just simulated a nation, played turn per turn, enjoying the journey rather than being obsessed about a score at the end of time. Like i play. Maybe thats why i suck and rarely win in Civ4, but at least i have a hell of alot of fun and immersion, playing to survive through the ages rather than predict every single tiny variable and strategically plan thousands of years into the future, gunpowder, nukes and beyond.

If i want to do that, if i want to play a game with a build order and tech order for optimal conquest, i'll play Starcraft 2.
I never played civ4 to play like that, nor did i buy civ5 to play like that.
But that is what civ5 appears to be, a game of less diplomacy, more RUSH RUSH WIN WIN SCORE SCORE, much inspiration from Civ Revolutions, it seems.
Which is a shame, i wish i'd known the game was like that before, i could just have kept playing SC2 instead and saved 60 bucks.

 

[Slowpoke]

I think the problem lies in what do YOU need to do to win each victory?

A) diplo: Kill as much as you can, get UN, bribe to win
B) dom: Kill as much as you can
C) culture: kill as much as you can, but puppet
D) Space: kill as much as you can for more science.


Until that changes, you can't fix the A.I. to make it optimize for its personality.

 

[testdummy653]

I think the problem lies in what do YOU need to do to win each victory?

A) diplo: Kill as much as you can, get UN, bribe to win
B) dom: Kill as much as you can
C) culture: kill as much as you can, but puppet
D) Space: kill as much as you can for more science.


Until that changes, you can't fix the A.I. to make it optimize for its personality.

I disagree, or maybe i agree...

The victory conditions need to be less dependent on killing/ their needs to be a penalty.
Like for cultural, make puppeted cities count as normal cities, so if your going for cultural... you need to raze them. Which means less production and gold for your army to expand and conquer.

Diplomatic, should be that your city-states are more mission dependant, and less money dependant.

Here's where i disagree... I'm not saying the AI need to be changed, I trying to explain why it does what it does. Not what needs to be changed.

 

[Haig]

I wish they just simulated a nation, played turn per turn, enjoying the journey rather than being obsessed about a score at the end of time. Like i play. Maybe thats why i suck and rarely win in Civ4, but at least i have a hell of alot of fun and immersion, playing to survive through the ages rather than predict every single tiny variable and strategically plan thousands of years into the future, gunpowder, nukes and beyond.

If i want to do that, if i want to play a game with a build order and tech order for optimal conquest, i'll play Starcraft 2.
I never played civ4 to play like that, nor did i buy civ5 to play like that.
But that is what civ5 appears to be, a game of less diplomacy, more RUSH RUSH WIN WIN SCORE SCORE, much inspiration from Civ Revolutions, it seems.
Which is a shame, i wish i'd known the game was like that before, i could just have kept playing SC2 instead and saved 60 bucks.

I know what you mean, I think Civ has two kind of players, the high score type and the ones like you who love the sort of storytelling/atmosphere style of having fun.
I definately play Civ IV the way you do.

I too think that giving A.I. play to win was a mistake from the V team.
Otherwise too it's too much of a turn based wargame instead of a Civilization game, even the people who love it seem to write just about fighting wars as it's about only thing to do(if you don't want to get bored to death going for cultural victory, which is a snorefest).
Nothing wrong about liking Civ V though, I don't mean that in my writing.

I remember reading before Civ V was out that if A.I. notices that it's victory goal is hard to reach it changes to another, and if it sees that player is approaching his goal, it will do everything to stop it. In reality I think this happens pretty rarely..

 

[testdummy653]

I remember reading before Civ V was out that if A.I. notices that it's victory goal is hard to reach it changes to another, and if it sees that player is approaching his goal, it will do everything to stop it. In reality I think this happens pretty rarely..

I agree it appears that it never happens.

But honestly, the war that was declared on you at turn 152 could of been a byproduct of the AI determining that you were about to beat them to victory that you weren't gonna attain till 425.

 

[SalmonSoil]

I think the main problem is that if the AI is doing anything amazing that would make us go out and buy a wedding ring so we can propose to it, the game is designed in such a way that we can't see it.

 

[fed1943]

Nice idea, but I think it lacks an important point.

A buddy player said once: we need to be careful not to lose, because we can't win if we lose.
So, the AI (and the player/s) always need to protect themselves, and to do it, something similar
to the conquest steps must be done (like to build and put enough troops, to get enough land and
so on).

Then, the conquest path is easier because something of it is already present.

 

[Clement]

Random Personalities.
But yeah.. I just wish the AI's did NOT have omniscient future planning minds, didnt view the game as a GAME, but rather did its part in suspending our disbelief (which is the developers part of the deal, our part is to immerse ourselves into the world and pretend its real. Sid Meier himself spoke about this for an hour at the GDC keynote, yet it seems the civ5 dev team completely disregarded it all).
Instead of, predicting gunpowder, every optimal path to a specific victory, and declaring war on everyone around in order to make the next 1000 years fit their conquest objective.

I wish they just simulated a nation, played turn per turn, enjoying the journey rather than being obsessed about a score at the end of time. Like i play. Maybe thats why i suck and rarely win in Civ4, but at least i have a hell of alot of fun and immersion, playing to survive through the ages rather than predict every single tiny variable and strategically plan thousands of years into the future, gunpowder, nukes and beyond.

If i want to do that, if i want to play a game with a build order and tech order for optimal conquest, i'll play Starcraft 2.
I never played civ4 to play like that, nor did i buy civ5 to play like that.
But that is what civ5 appears to be, a game of less diplomacy, more RUSH RUSH WIN WIN SCORE SCORE, much inspiration from Civ Revolutions, it seems.
Which is a shame, i wish i'd known the game was like that before, i could just have kept playing SC2 instead and saved 60 bucks.

I really agree with you, my first Civ game was Civ2, since that game there has always been a strong "roleplaying" element if it must be called that way, to every iteration in the series, i don't understand why the "roleplay" label is being attached now since it's been in every single Civ except for this one, the new concept here after all is "playing to win".

Players like you and me that play for immersion and the emergent gameplay that the Civ series used to offer have been left out this time, and yes, i too wish i had known about this design decision before i purchased, the Civ series was always about so much more than just war, i wait to see what the next patch will bring.

Random Personalities.
But yeah.. I just wish the AI's did NOT have omniscient future planning minds, didnt view the game as a GAME, but rather did its part in suspending our disbelief (which is the developers part of the deal, our part is to immerse ourselves into the world and pretend its real.

Also this bit, this captures so well how i feel about Civ 5, better than i could have said myself, the arbitrary domination and diplomatic victories mean that some cities are made of "magic bricks" that cannot be razed, i just can't suspend my disbelief enough for that.

 

[MkLh]

This is a good point. When they choose to dumb down diplomacy, personalities etc. and make AIs human like aggressive kill bots, the also make any other win condition than domination/conquest unreachable to AI. Human may still pursue space and culture because human is smarter, but if some AI tries something like culture against equally "smart" AIs, it will be stomped. In MP, culture and space are not very popular either.

But you can also ask, is AI able to win domination or conquest either. In CivIV, I never saw AI winning those conditions (I saw it winning by culture, diplo and space). Of course, domination is now made a lot easier. I wonder if the reason for that laughable domination condition is to offer AI at least one possible way to win.

 

[Derax]

The developers should just make it so that being at war costs more mainteance than being in peace.

To this moment I actually can comprehent the Ai' intentions because If I was in their situation I would do it to(give a DOC) ...

A solution maybe would be the introduction of historical disputes just like in emprie earth which then reduces the probability of war.

 

[ecki2k]

How about buying City States for the Bonuses (food, culture,units) but for them to vote for you, you need to fulfill at least 3 missions for them?

EDIT: Sorry, doesn´t belong here. But I agree with OP

 

[wurstburst]

It's possible that what the OP says is true. Makes some sense from what I have seen.

However, at the root, it doesn't matter, because the AI is not good enough at battle, nor smart enough in planning to play a good defensive game with sufficient defenders to go for the other non-domination type of wins.

There was a thread (great thread!) where someone went with a single city, diety, always war against all the AIs. They were able to play defensively and hold off all the other civs long enough to get to artillery. 3 citadels, ultra promoted trebucket one-shotting modern units, defense, defense, defense. The AI can't do this. But, there should be some effort put into making it a better defender (both when attacking and defending!). That would go a long way to making the military-minded civs less successful and allow a culture or diplo minded civ to thrive better.

Once that happens, the programmers could then focus on making the AI better at getting to the other victory conditions.

 

[ProximaCentauri]

I think the problem lies in what do YOU need to do to win each victory?

A) diplo: Kill as much as you can, get UN, bribe to win
B) dom: Kill as much as you can
C) culture: kill as much as you can, but puppet
D) Space: kill as much as you can for more science.


Until that changes, you can't fix the A.I. to make it optimize for its personality.

Totally agree with that.
I find myself just founding 1 to 3 cities in every game and warmongering, gettin a lot a puppets. They generate high amounts of science, allowing for science victory, the low number of founded cities and the medium culture generation of the puppets is still very good for culture victory (at some point of warmongering i get new policies every 4-5 turns until I have 5 full branches, despite the fact I often spend 2 policies in the first branch then leave it alone), and of course if I'm not bored I can continue the world conquest. As for the diplo victory... just too easy to spend a word on it.

 

[Chiyo]

I know what you mean, I think Civ has two kind of players, the high score type and the ones like you who love the sort of storytelling/atmosphere style of having fun.
I definately play Civ IV the way you do.

This is exactly how I feel about the new CIV. It is not a game of "built a civilization to stand the test of time" but more like "built a civilization to win the Civ 5 match". I really hope the developers recognize that they have taken a step in the wrong direction and manage to turn it around. I mean, after nearly 20 years of playing Civ this cannot be the end for me?

 

[testdummy653]

Nice idea, but I think it lacks an important point.

A buddy player said once: we need to be careful not to lose, because we can't win if we lose.
So, the AI (and the player/s) always need to protect themselves, and to do it, something similar
to the conquest steps must be done (like to build and put enough troops, to get enough land and
so on).

Then, the conquest path is easier because something of it is already present.

I mentioned this when i talked about India, and her cultural victory. Yes, India will have a military, but it will not be as big as Siam, who has a big army, and need to conquer to win. India could win, but it unlikely if she gets a DoW, and loses production on cultural buildings, to push out military for defense.

But yes military is one of the keys to survival, in the real world and in CiV.

 

[Arsenic Steel]

I mentioned this when i talked about India, and her cultural victory. Yes, India will have a military, but it will not be as big as Siam, who has a big army, and need to conquer to win. India could win, but it unlikely if she gets a DoW, and loses production on cultural buildings, to push out military for defense.

But yes military is one of the keys to survival, in the real world and in CiV.

Having a military does not effect a civs SP or cultural output. There is a limit to the number a cultural buildings and Wonders that can be built in any given era.

The flaw in your logic is that you infer having a larger military force means that a civ is going to conquer others, it could be present solely for defense. Sadly that flaw is true in civ V and part of what is wrong with the AI's victory conditions.

 

[Clement]

Having a military does not effect a civs SP or cultural output. There is a limit to the number a cultural buildings and Wonders that can be built in any given era.

The flaw in your logic is that you infer having a larger military force means that a civ is going to conquer others, it could be present solely for defense. Sadly that flaw is true in civ V and part of what is wrong with the AI's victory conditions.

you make a good point, if the AI is subtle enough to own a large military but only use it for defence, i have'nt seen it yet.

 

[Zogar]

Are there really people saying that the AI can't win another way because of the game rules ?

How come this 30+ city civ, ALONE on its continent, with double my income, land, tech and 5 times my army cannot win diplomatically or space ship and let me win culturally ? There's no excuse for this. I don't know how anyone could blame the game rules and not the AI developpers on this -_-.

 

[Drawmeus]

Random question -

If you have a powerful military, what reasons should be in place where you WOULDN'T want to conquer that neighbor, even if you aren't setting out to conquer the world?