Broken warfaring?

[Terxpahseyton]

I noticed something very strange and frustrating in my recent game. Basically, the AI does not use its armies.
For instance there is Persia, the by far strongest power. With an army five times the power of India. And with a shared border with India. And hundreds of years of war. with India. But no conquests. Zero.
I finally open the world-builder and noticed, that Persia had some units just sitting there at the border and that was it.
Another example is Greece and Hittites.
Greece is the weakest civ of them, Hittites one of the stronges. Greece has a city which is cut off from the rest of Greece and encircled by Hittite cities and territory. And Greece and Hittites pretty much have had constant war since forever. No conquest of that city....

In fact, after a few city conquests in the early game, war has seized to actually be carried out in the entire world, it seems.

You may think that this may be due to the lack of bombard units on the one hand and cities which can not be attacked without reducing the defense on the other - but I gave battle rams etc. bombard capabilites (a tiny bit of xml-modding on my part), so that is not it, I think.

Also, before this game, which is a by now a bit dated svn-version, I played another game with the "patched" (found the patch in some thread) 36-version and there was quit a bit more conquering going on. The Zulus even were on a road to capture the entirety of Africa.

That is such a major thing that I should not be the only one experiencing this. So I am wondering: is there an awareness of this problem? It really breaks the game for me right now, which is a huge shame since I had such a great time playing this Giant Earth Map scenario after I gave the AI ridiculous boni and finally got some suspense going.

 

[Keyalha]

Does the AI recognize your alterations? If the AI still has their old build order preferences nothing will change on the AI end if you do not alter stuff there.

IIRC all the Civ 4 AI does is using weighted preferences on units. I do not know how extensive the AI has been rewritten in this mod so take what I say with a grain of salt.

In short: I think the AI does not even recognize your change.

 

[Terxpahseyton]

The AI does extensively build battle rams, in my recent experience. With or without the change. And I would assume that once the AI has the option to bombard a city with units within its stack to reduce ti defense, it will do so without needing further cues (I'd think). A thought that just occured to me: when you can not attack a city until you have reduced its defenses - maybe the ai proccesses this information as simply "I can not attack this city - period" and hence mounts no effort whatsoever?
In my earlier game, all the conquering happened at a fairly early technological stage, as well. Possibly at a stage where the building which forces you to reduce defense to be able to attack wasn't yet there.

Btw - I think it is a very weird design choice to introduce this feature of unattackable cities well before siege weapons exist. I suppose the point is to slow down empire expansion. That is a bit of a too drastic measure for my taste, though.

 

[Thunderbrd]

I was playing a test game and despite being fairly knowledgeable regarding the AI, was shocked to be attacked so early into the game by a civ that I'd met but didn't even know where they were coming from. They did a fairly good job coming at me too... it took quite a bit of building up my forces to counter them. They didn't quite bring enough in to take me down but it certainly hindered me quite a bit. So I really think the situations are so complex that it varies a great deal from one game to the next how effective they are.

 

[Keyalha]

The AI does extensively build battle rams, in my recent experience. With or without the change. And I would assume that once the AI has the option to bombard a city with units within its stack to reduce ti defense, it will do so without needing further cues (I'd think). A thought that just occured to me: when you can not attack a city until you have reduced its defenses - maybe the ai proccesses this information as simply "I can not attack this city - period" and hence mounts no effort whatsoever?
In my earlier game, all the conquering happened at a fairly early technological stage, as well. Possibly at a stage where the building which forces you to reduce defense to be able to attack wasn't yet there.

Btw - I think it is a very weird design choice to introduce this feature of unattackable cities well before siege weapons exist. I suppose the point is to slow down empire expansion. That is a bit of a too drastic measure for my taste, though.

That is the point, the AI does not have the option as far as i know. The AI acts with the units as told like a child would. I.E This unit is for bombardment and such.

As said all this is said with how the vanilla civ 4 ai works without knowing how extensive the rewrite of the AI is.

As a general rule of thumb you can and will break things if you modify things.

 

[JosEPh_II]

<snip>

Btw - I think it is a very weird design choice to introduce this feature of unattackable cities well before siege weapons exist. I suppose the point is to slow down empire expansion. That is a bit of a too drastic measure for my taste, though.

Here is a point I'm in agreement with. Had more than 1 discussion over the past 2+ years over it. But the reasoning given for making it this way was to stop the player from Steamrolling the AI.

Unfortunately the consequences are that the AI will start a war but once it's 1st wave of attackers don't take a Players city down it then goes into massive Ram build up. But...it is currently Not as bad as it was. Making rams 1 pt of str weaker and upping spearman 1 str and Axe 2 str has helped the AI sustain their attacks And more importantly cut back on Ram build up.

But once the AI in a war is put on the defense it goes back to making stacks of Rams with 1 or 2 regular fighting units mixed in. It stops making stacks of Real fighting units. The weighting for Ram usage is still too much imhpo and needs reduced in the code further. But there are other factors that play in as well so it's not as straight forward as you might think.

JosEPh

 

[Terxpahseyton]

I was playing a test game and despite being fairly knowledgeable regarding the AI, was shocked to be attacked so early into the game by a civ that I'd met but didn't even know where they were coming from. They did a fairly good job coming at me too... it took quite a bit of building up my forces to counter them. They didn't quite bring enough in to take me down but it certainly hindered me quite a bit. So I really think the situations are so complex that it varies a great deal from one game to the next how effective they are.

As said attacks early in the game I have also seen plenty. In the first half or so of the ancient era Persia quickly conquered all of Babylon, for example. And I was like "wow this is great", because for one I appreciate the action and for another it means an AI player is getting stronger and I basically want AI players to be powerful enough that I have to seriously worry how I will compete with them on the long run.
But then towards the end of the ancient era conquests just stopped to happen even though a lot of war is going on and a lot of units are at the AI's disposable. To stress: This happened on a global scale and I am playing the GEM scenario with more than 30 civilizations. So that is hardly a fluke, but clearly appears as a deep-rooted systematic issue, don't you think?

I think I will disable the function of defenses which have to be reduced to be able to attack and see if that leads to any change.

As said all this is said with how the vanilla civ 4 ai works without knowing how extensive the rewrite of the AI is.

I am not experienced with telling the AI what to do with units. I'll have to see weather there is an easy xml-fix, I think there are some sorts of strategy-tags in the xml-unitinfos-file. Maybe I just have to change those. Insert the strategy-entries of catapults or something.

 

[Thunderbrd]

Combat AI is due for a number of major overhauls. I'm not going to be throwing the babies out with the bathwater as there are brilliant sequences in the overall picture, but a complete analysis and restructuring, which could take up to a year or more, is certainly in order. There is, simply put, no greater complexity (aside from maybe trying to understand multi-threading issues and OOS issues) in the code than the AI. There is absolutely nothing simple nor straightforward about it.

This version will not address large scale AI issues like these first. Instead, I will be seeking to greatly improve their support unit activities, stealth unit activities, and stealth response activities. Next version will get the full workover of the land based military units and their AIs.