Questions About Civs' Aggression

Ozymandias

In Terra Fantasia
Supporter
Joined
Nov 5, 2001
Messages
10,828
Location
The lone and level sands
The "ferocity" hard-coded into Rome is fairly well known (to paraphrase @AnthonyBoscia, "That's why I always use 'Red Rome' - :satan: - for Russia.")

As hindsight always seems to be 20:20, going back as far as my early testing for what units the AI would build next, Egypt seems well, pretty lame, and Greece not far behind (for my tests, I simply used the first three Civs.)

After a recent conversation with @Vuldacon about Egypt, I'm very curious to hear about anyone else's experiences along these lines.

Cheers,
:hmm:z
 
I have always viewed Egypt as a non-aggressive civilization. As for the Greeks, I periodically play them, and if I am not, they normally are not in the game. I will have to do some experimenting before saying more on them. However, the Vikings do tend to go Berserk.
 
I have always viewed Egypt as a non-aggressive civilization. As for the Greeks, I periodically play them, and if I am not, they normally are not in the game. I will have to do some experimenting before saying more on them.
TY, once again!
However, the Vikings do tend to go Berserk.
:viking: (And there I thought I'd never have an occasion to use that particular emoji!)


:D
 
As Chieftess's signature says, "Never trust an Aztec with nukes!" There are definitely different aggression levels. I've long advocated for the aggression slider on the "Civilization" tab in the editor(s) being by far the most important, if not the only, factor in AI aggression.

That said, after the recent thread in Civ3 General about Gandhi, I'll grant that military strength may be another factor. The Vikings may be more aggressive than other civs of the same aggression setting (4/5 in an unmodded game) during the age of Berserkers, because their army will be proportionally stronger due to Berserks, and the AI will likely factor that in when deciding whether to start a war.

How much impact the aggressiveness setting has in practice is debatable due to the amount of data still adding up to "plurals of anecdotes". Someone could probably do an analysis of the Game of the Month archives, and see how much of a statistical variance there is between civs in terms of city conquest, when the civ starts on a different continent than the human. It's not a perfect data set (notably I don't think the game records who started wars, and there still are human players in the game), but it's the best I've thought of in terms of having enough games to add up to being statistically significant.

But short of some statistical evidence showing otherwise, I'm still of the opinion that aggression levels are the primary factor, with relative power (potentially impacted by unique units, or civ traits such as militaristic) as the secondary factor. I do not believe there are any hidden factors. Rome is 4/5 aggressiveness, so is Russia. Rome has an early and relatively powerful unique unit, so they may seem more aggressive due to being relatively more powerful in the early game (and consequently in the later game if they win early wars), but I don't believe the AI for Rome is inherently programmed to be more aggressive than Russia, Persia, or any other 4/5-aggressiveness civ. If you gave them the same traits and same unique units as Russia, and ran 1000 AI-only games, I would expect the difference in aggressiveness to be statistically insignificant.
 
Over all, I believe there are different Aggression Levels programed for each Civilization other than the Aggression Slider Bar in the Biq.

Another Factor I have noticed (Perhaps the Strongest) is when a Civilization gains Power from Many Strong Units, they will go to War. An example is when India gains War Elephants and has many of them, you can depend on them to Attack.

I agree with Quintillus in that Unique Units and Civ Traits add additional factors to Civ Aggression.
 
To clarify, I chose the first slot for the Soviet Union in my scenario because they are the predominant civ and the game is built around them. The prospect of them occupying the civ slot of maximum aggressiveness was meant as a joke. In my other mod Rome still retains this spot as Italia.

I'm inclined to agree with Quint as I've never seen any evidence outside anecdotes that certain civs have a hidden aggression factor. This especially seems to come to light in mods where the civs bear no relation to historical civs. I think the Roman rabbits foot is one of those long running civ superstitions, although I'm willing to concede being wrong if we got some hard evidence. Or maybe we prefer to have a little magic still hidden under the surface.
 
Not sure what Evidence would be certain but for example, I set up a Mod where India's Aggression was set to the least in the Biq.

Had Good relations with India.

There was no behavioral difference because when they gained and had many War Elephants, they surprise attacked me... same as usual.
 
The "Build often" flag matters too besides the aggression slider. If a civ has a superiority in attack stats (bombard is also factored in), then it's more likely to prey upon those. If "Build often" is flagged for Offensive Units, the civ is more likely to have more of them during peace time.

Rome's stock game aggression is only 4/5 in stock game. Bismark, Temujin and Shaka are at 5/5 and they too start with a stack of archers. These 3 are the most likely to kill you outright on higher difficulties and there's no counter play.
 
I will do some testing as well with respect to India and aggression.
Oh, the notorious Gandi! :shake: I've reading o/l articles dismissing this well-known "myth."

When I joined the "general" "All Civs FB Group," the "Say Sesame And Enter" question concerned Said Leader's "Special Characteristic."

I answered, "He likes nukes," and the algorithm let me right in.
 
From doing lots of AI only tests on identical earth map starting positions and a bunch of actual games I'd throw in a few subjective observations based on the following aggression rankings:

5 = +2 (Germany, Zulu, Mongols)
4 = +1 (9 Civs)
3 = 0 (6 Civs)
2 = -1 (China, Iroquois, Carthage, Korea)
1 = -1 (France, India)


1) its almost like the Germans and Mongols attract war in a way that leads to their own demise. Perhaps their irritability and aggression translates into threatening other AI to an above average extent, poisoning relations and encouraging a dog pile against themselves? I do not experience this with Zulus and would attribute this to them having a defensive early game UU (see point above about prioritising offensive units) and because they are stuck on a peninsula in southern Africa and so have a lot less scope for war as they are mingling a lot less with others. In general, these 3 Civs never threaten victory in any of my games due to their apparent terrible diplomacy skills. It makes me miss Civ 1 days when high aggression Stalin, Shaka and Ghengis could be relied upon to be the toughest foes in the early to mid game. Whereas in Civ3 they high aggression ones are often the first to die. It might be different if they had an early game offensive UU to compliment that aggression, but they don't.

2) loads of people talk about the Aztecs as aggressive. I've rarely seen it and I've never been concerned by their UU because it is so easy and low cost to avoid war in the early game. There also seems to be no sneak attacks in the early game as the AI doesnt seem to consider this until all land has been colonised and by that time you'll easily repel any Jaguar Warriors. I'd maybe double their cost but give them an attack of 2.

3) Civs like China, Iroquois and Persia absolutely thrive on a consistent basis, and only one of these is Agricultural. Their lack of aggression typically means they are late to join a dog pile and so enter wars with max offensive capacity at a time when their foe has lost its military power from 5+ turns of war. This is a devastating combination and in my experience leads to far more efficient warmongering from lower aggression Civs. I personally dislike this as it seems counterintuitive. Persia in particular is an absolute god at this. It can turtle peacefully, tech at a good rate despite being a smallish Civ (even more so than other scientific civs), then use a tech advantage to efficiently dogpile from late medieval onwards. The Persians and Chinese can come from nowhere to be a contender (whereas Iroquois never fall behind because of Agriculture).

4) High aggression Civs are more prone to fall behind on tech due to prioritisation and getting sidetracked into wars. I also don't like that and it can make their aggression really counterproductive.

5) the AI, as per the link above, does seem to be pre-disposed to hate the leader on score. So dog piles can occur on lower aggression Civs in these circumstances.

6) I would let the Ottoman and Koreans babysit my children. I find them more passive than France and even India. Thinking about it more, I find the Scientific trait may coincide with peacefulness but can't imagine why that would be.

7) no country has reportedly ever won more battles than France since the dawn of documented civilisation. Why on earth are they lowest aggression?


I don't do mods, so my experience and knowledge is limited. But I'd argue the aggression in Civ3 isnt it's best feature. Personally if I was modding I would:

A) marry high aggression to those Civs with early game offensive UUs (e.g. Rome, Persia, Iroquois, although with Agriculture this could be a lethal combo) to try and create a synergy.

B) reduce aggression levels for those with late game UUs or non-offensive UUs. These guys will still get aggressive but you are raising the chances of that aggression coinciding with them having a power spike rather than them looking like an infant throwing a tantrum who then gets dogpiled.

C) prioritise worker building for high aggression Civs with early game offensive UUs. There is nothing worse than being a warmonger that hasn't hooked iron or horses in your territory. Germany is particularly guilty of doing this and being first to be destroyed.

D) the Flintlock patch letting AI use armies is probably a huge variable and arguably the existing flaw that massively undermines high aggression AIs and prevents them snowballing in a way that might help them withstand being victim of the bad sentiment that their above average aggressive diplomacy may cause. I'd argue aggression is fundamentally broken if the AI cannot use military leaders and create armies. This is probably next on my tinker list to see if it makes aggressive militaristic Civs as terrifying as they should be, rather than deeply underwhelming.
 
This discussion of civilization aggression levels has gotten me to examine all of the civilizations for aggression. The ones that need adjusting in my view for Conquest are as follows.

France: The French have been trying to extend their territory since the time of Charlesmagne, and it is only since the defeat by Prussia in 1870 that they have pulled in their horns. Therefore, I have boosted them to a 4 for aggression.

Iroquois: The Iroquois were in the process of building up their own empire when the Europeans arrived, by military conquest. They were allied with the British in both the American Revolution and the War of 1812, and the British had a difficult time trying to control them. I raised their aggression level to 4.

Spain: The Spanish spent centuries trying to throw the Moors out of Spain, and when they finally succeeded in 1492, they continued to try to expand their holdings in Europe, as well as taking over large areas of the New World and in Asia. I would raise them to a 4.

Ottomans: The Ottoman Empire was based on conquest, especially in Europe and continual struggles with Persia. The expansion attempts finally stopped after the 2nd Siege of Vienna in 1683. I would raise them to at least a 4, if not 5.

Inca: They had built an empire running for over a thousand miles through the Andes. I would rate them as a 4 long with the Iroquois and Aztecs. I would be inclined to give the Aztecs a 5 because of their continued need for prisoners for human sacrifice.
 
For game balance I lowered Germany to a 4 (to give them half a chance to live long enough to use the panzers) and raised Egypt to a 4 (to encourage use of their UU).

For historical interpretation (I'm no expert, but my novice interpretation of an empire's willingness to attack an external rival with comparable technology) I lowered Babylon to a 2, Japan to a 3, Maya to 2, Sumeria to 1. For the same reasons I raised France and Ottoman's to 4 and the Hittites to a 5.

I'm hoping removing the 'attacking unit' AI flag for Impi and Numidean Mercs will make the Zulu and Carthage less rubbish in my games (so I've held off on tweaking their aggression levels yet). There is nothing sadder than seeing a stack of AI Carthage UUs used offensively.

I've only played two games since the tweaks but it was nice to see my newly aggressive Ottomans attack me (first time ever), as well as my minimum aggression Sumerians steamroller an entire continent in the early game and my newly pacified Japanese assault me by literally throwing a bunch of warriors at my basic chariot hordes, before Japan had even settled half it's available land. This is all on default gamewide aggression level so it just shows there are plenty of variables behind the AI decisions around war beyond the simple 1 to 5 score. I guess the aggression score could be thought of as a slider for the willingness of the AI to start a sub-optimum war without any provocation, but even an aggression score of 1 will be as aggressive as heck if they calculate they have a significant upper hand.
 
I guess the aggression score could be thought of as a slider for the willingness of the AI to start a sub-optimum war without any provocation, but even an aggression score of 1 will be as aggressive as heck if they calculate they have a significant upper hand.
I agree and especially concerning the AI calculating they are much stronger :yup:
 
I don't think I've got aggression sorted at all and don't like some of my changes. Although it does seem to me like aggression of 2 or 3 (on average) is the AI sweet spot in terms of avoiding premature wars but having just enough aggression to capitalise ruthlessly when the AI gets an advantageous position. Having Japan at 3, they were ruthless in the medieval era onwards and destroyed my perception that 'militaristic' was more powerful as an early game trait.

What I'll be trying now is thinking of the aggression level in terms of 'how early in the game will this AI (on average) pick a fight). So I'm picking the aggression level based on my perception of the Civ historically then:

+2 if UU is available in first column of tech
+1 if UU is in Ancient Era
0 if UU is in Medieval Era (excluding Military Tradition)
-1 if UU is Military Tradition or Industrial Era
-2 if UU is Modern Era

The exceptions are Civs I want to nerf (pretty much all Agricultural I don't want having early game aggression as that would synergise with their trait too much) or buff (Mongol and Rome not made a 5 for aggression because I want to increase the chances of them being a long term threat).
 
I'd like to note here that with Flintlock's patch you can nerf Agricultural civs by making their extra food from city centers be applicable for the despotism tile penalty as well, so they'd only get their bonus once they switch governments. That could already be all the nerf you need for their early game aggression.
 
Top Bottom