AI stays too small, reason?

[Furycrab]

Theres a question that I cant answer myself, please enlight me:

The situation is as follows: Theres a big continent. I am in one corner, AIs are spread over the rest. Theres still lots of free space.

Then, the AIs build a few citys each (3?) and stop growing. They declare war on each other over huge distances and sign peace a while later.

They totaly stop growing alltogether and the space between them stays unused. Even with my building/culture-oriented play I outgrow them by far, to a point where its boring. It happeneds way too often

Is this related to difficulty? Is there a max size they aim for related to difficulty?

What type of map are you playing?

Please don't say Earth.

 

[BSPollux]

900hs of playtime say its has been most maptypes ;-)

Well that specific example was arborea large (continents) and its produced a huge continent with me and the arabs at the very south and persia, russia and sweden at the corners of a huge northern landmass.

But I am very used to see that behavior on all kinds of maps. thats why im asking you all here if theres a trick to avoid it.

 

[hobbsyoyo]

So Tachii said its related to nation: Some settle more then others.

Putmalk speaks about a treshold of exansion, related to gametime?

Is that common knowledge about the AIs behaviour? that each has something like a given timespan of initian expansion and if they get interrupted they just stop?

I believe all ideas on how/why this happens that have been posted on this thread are conjecture and speculation, including my own. I have not seen a dev or anyone who has knowledge of the dll files comment on this issue.

 

[Furycrab]

Fair enough, I've seen it before... The problem is two-fold.


1) When the AI wants to play for a peaceful/OCC type victory. It's cool that they do, however very unlikely they will win unless you crank it up to immortal or Deity.


2) The AI is a bit uncertain with how to proceed and expand when it's tied up in a war. More precisely, doesn't know how to recognize that it can't win the war and should just expand/build more infrastructure/get more tech till it can.

In both situations I've seen improvements in G&K, but obviously not perfect.
I've still been in wars where the AI can't recognize that it would practically need airplanes to get past the terrain I can easily defend.

 

[Jabberwockxeno]

Never seen this happen.

 

[Peacemongerer]

I find that the AI who stay at one city do so trying to get a cultural victory (They almost always adopt Piety). Those that expand often get halted when they lose a settler. I had a game where Ramses stayed at one city until I took a barb camp and gave him back his settler (I hadn't been paying attention and thought it was a worker). This was like 600 AD.

He ended the game with 12 cities when I won in 1923 and appeared to be on his way to more had I not won nor he ran out of room.

 

[stii]

I had a Civ on deity stick at two cities for most of the game which was very weird. He was near me so there was plenty of room, but it certainly made the game a lot easier

 

[nokmirt]

The AI also does not take into account the massive gold pile they are sitting on that they could use to improve their new undeveloped cities.

Perhaps raising some of the leaders flavor values via xml may solve this. Also, in the GlobalDefines.xml raising things like units for city defense, naval, and other things, may help the AI spend more money. If you want the AI to do more tile improvement raise the value for that flavor in the leaders xml. Some civs have very low flavor values in several departments, which they do not improve a great deal. This causes them to hoard a huge supply of gold.

For the most part in GnK, I have seen the AI do far more tile improvemnet than in vanilla. I am always pillaging tons of tiles. I don't remember doing that quite so much in vanilla.

<Row>
			<LeaderType>LEADER_ATTILA</LeaderType>
			<FlavorType>FLAVOR_TILE_IMPROVEMENT</FlavorType>
			<Flavor>5</Flavor>
		</Row>

Attila is not a big tile improver, which realistically how he should not be. However, if you changed the flavor value to say 7 or 8. There should be marked improvement in that department.

Now look at these, Attila is perhaps the most belligerent civ. Let's compare him to Monty.

<Row>
			<LeaderType>LEADER_ATTILA</LeaderType>
			<MajorCivApproachType>MAJOR_CIV_APPROACH_WAR</MajorCivApproachType>
			<Bias>8</Bias>
		</Row>
		<Row>
			<LeaderType>LEADER_ATTILA</LeaderType>
			<MajorCivApproachType>MAJOR_CIV_APPROACH_HOSTILE</MajorCivApproachType>
			<Bias>8</Bias>
		</Row>
		<Row>
			<LeaderType>LEADER_ATTILA</LeaderType>
			<MajorCivApproachType>MAJOR_CIV_APPROACH_DECEPTIVE</MajorCivApproachType>
			<Bias>8</Bias>
		</Row>

<Row>
			<LeaderType>LEADER_MONTEZUMA</LeaderType>
			<MajorCivApproachType>MAJOR_CIV_APPROACH_WAR</MajorCivApproachType>
			<Bias>8</Bias>
		</Row>
		<Row>
			<LeaderType>LEADER_MONTEZUMA</LeaderType>
			<MajorCivApproachType>MAJOR_CIV_APPROACH_HOSTILE</MajorCivApproachType>
			<Bias>6</Bias>
		</Row>
		<Row>
			<LeaderType>LEADER_MONTEZUMA</LeaderType>
			<MajorCivApproachType>MAJOR_CIV_APPROACH_DECEPTIVE</MajorCivApproachType>
			<Bias>7</Bias>
		</Row>

Yes, Attila is not a very nice fella to deal with.

 

[Putmalk]

Okay, allow me to explain.

The AI is split into different sections. Grand Strategy, Militaristic, Economic, and Diplomatic.

Grand Strategy determines the overall victory the AI will be pursuing. Each strategy has a flavor associated to it, so when the AI chooses to embark on this strategy, they will focus on certain things (Cultural won't expand as much, you'd hope...).

Militaristic is obvious, the AI builds units depending on what technology they have researched.

Economic is the one you're looking for. The AI has an EARLY_EXPANSION strategy, as well as an EXPAND_TO_OTHER_CONTINENTS strategy. The expand to other continents doesn't fire until TECH_OPTICS is reached. The early expansion has a time after which it expires (I may be slightly wrong, I have to check the files) but there has to be a definition which states when early expansion stops.

So, my theory is that the AI loses its settlers early, and then the early expansion strategy expires and they stop building more.

I do not have access to the coding, but it's very clear to every modder that the AI builds units based off of various FLAVOR thresholds. FLAVOR_EXPANSION is the relative one here. These flavors are adjusted based off of the situation (the AI loses any desire to expand when they're being seriously threatened for city defense, as focus shifts to supporting their cities). I think to correct the problem the AI should bring an additional unit to protect their settlers (force that unit to be within a 3 tile radius of the Settler, and have it attack anything that attacks the Settler's guard).

This information is coming from my various time peeking in the AI XML files. My knowledge is not perfect, but for this I'm pretty sure of.

 

[Txurce]

So, my theory is that the AI loses its settlers early, and then the early expansion strategy expires and they stop building more.

I do not have access to the coding, but it's very clear to every modder that the AI builds units based off of various FLAVOR thresholds. FLAVOR_EXPANSION is the relative one here. These flavors are adjusted based off of the situation (the AI loses any desire to expand when they're being seriously threatened for city defense, as focus shifts to supporting their cities). So it is my theory that they are losing the city settlers. I think to correct the problem the AI should bring an additional unit to protect their settlers (force that unit to be within a 3 tile radius of the Settler, and have it attack anything that attacks the Settler's guard).

This information is coming from my various time peeking in the AI XML files. My knowledge is not perfect, but for this I'm pretty sure of.

Agreed top to bottom. In another thread I proposed that having the AI always escort settlers would go a long way toward making it more competitive. I also know from playing with mods that even with the tools available now, AI expansion can be noticeably adjusted. (The AI has always had a late-stage expansion, from vanilla on.)

 

[bcaiko]

I find that the AI who stay at one city do so trying to get a cultural victory (They almost always adopt Piety).

And they almost always get swallowed up by someone going for Domination Victory (or just being a jerk). The AI should be programmed to avoid OOC. It's a "challenge" for a reason...it's not a good strategy. Even for a Culture game.

 

[fredmeister]

This information is coming from my various time peeking in the AI XML files. My knowledge is not perfect, but for this I'm pretty sure of.

I started a new marathon game with no barbarians (no chance for settler loss) and the problem persists. I was well into the medieval period and Persia had 2 cities, Germany 3. So I guess it's not related to losing Settlers.

 

[Zanarkin]

I started a new marathon game with no barbarians (no chance for settler loss) and the problem persists. I was well into the medieval period and Persia had 2 cities, Germany 3. So I guess it's not related to losing Settlers.

The civ will spread their wings - it's just not a priority at that point for them. Playing difficulty six, 3/4 of the civs spread asap and the 1/4 stayed small until the twentieth century. By the twenty first century the only civs that had not expanded were those constantly getting ganged up on by other civs and on life support.

Some civs are just late bloomers and you guys finish the game before reaching the 20th century. I don't even consider the game 'started' until I hit the modern era on my marathon games so I have seen countless times the late bloomers bloom.

 

[Putmalk]

Not all civilizations will expand. If they are going for Culture they will tend to stay small.

 

[nokmirt]

I started a new marathon game with no barbarians (no chance for settler loss) and the problem persists. I was well into the medieval period and Persia had 2 cities, Germany 3. So I guess it's not related to losing Settlers.

This is interesting. Also, the flavor values are the same for the most part for leaders in GnK, as they were for vanilla. Even the AI economic strategy values in the xml code are the same, as far as i can tell anyway. So, why are civs not building cities like they used too?

What is weird is in my 'Into the Renaissance' game I am playing, Spain oddly does not have the Iberian peninsula filled up with cities. This far into the game it should. There are about 100 turns left. I sent a scout over there from North Africa and there is tons of land in southern Spain. I traveled north a bit more and found a bunch of barb camps. It is really strange. I realize the map is huge, but that is ridiculous. Maybe Spain is much larger than I expected.

What I would like to know is, if you code the AI civs not to expand, why make a scenario with a huge map, and no chance of the civs building enough cities for rivals to conquer in the allotted number of turns? In this case 200.

I have my army ready to invade Spain, but after a trek for some distance inland I will just be invading barb camps. If I wanted to fight barbs there are plenty in North Africa, and therefore no need to go to Spain. My scout will carry on, perhaps someday before he gets too old, he'll finally make it to Madrid.

 

[danieladler]

Play emperor, and if you can stand the super corruptness and over-obvious cheating, play immortal.

 

[smallfish]

Immortal is alright

I'm still trying to find that sweet spot between emperor and 'mortal, and the closest I've gotten is playing on a Large Map with Quick Speed Emperor.

 

[BSPollux]

This is interesting. Also, the flavor values are the same for the most part for leaders in GnK, as they were for vanilla. Even the AI economic strategy values in the xml code are the same, as far as i can tell anyway. So, why are civs not building cities like they used too?

Maybe they spent time on religion and that messes up the grand plan?

Plus the free settler from liberty that comes way later now. or the worker, if you select it the over way round, whatever

 

[hobbsyoyo]

I started a new marathon game with no barbarians (no chance for settler loss) and the problem persists. I was well into the medieval period and Persia had 2 cities, Germany 3. So I guess it's not related to losing Settlers.

It may not be related to losing settlers, but I bet CS's play a part in it. I've noticed the AI trying to rush CS's early game, and when they lose, they stop expanding. It seams they spent all their efforts and resources on the rush and can't switch gears back to expansion.

Someone mentioned the free settler coming later in the Liberty tree. Wouldn't that be hilarious if that was the problem? The AI can't get that settler early and loses all the momentum that that early boost gives it and can't cope.

The more I think about it, I actually truly think this may be part of it. Why did they change the tree around anyway? Gameplay balance?

 

[fredmeister]

Next time I will try to play on marathon with:

- No barbarians
- Difficulty 6
- No city states
- No non-military victory conditions