AIs and the Art of War

I'm genuinely glad you're "having fun", but please tell me that you're aware that both of those (runaway money and GA length) are serious bugs... Is anything being done about them?

Depends a bit on difficulty level. And a look at the minimap shows that he controlls quite a large chunk of the world, so lots of cities to produce GP.
 
I'm genuinely glad you're "having fun", but please tell me that you're aware that both of those (runaway money and GA length) are serious bugs... Is anything being done about them?

Depends a bit on difficulty level. And a look at the minimap shows that he controls quite a large chunk of the world, so lots of cities to produce GP.

Yeah both of these "oddities" were fix (well i hope so at least, by Koshling). He has fined tuned the cumulative money problems and i have fixed the GP stuff (I hope).

This is an OLD game, so alot of stuff is running different here. I am only actually getting around 27-39 thousand dollars per turn now.
 
Teach AI to gift units among each other

Inspirated by current situation in Syria where US gives weapons to rebels and Russia gives weapon to Assad goverment I noticed funny situation. This two countries (USA and Russia) are not at war but they fight against each other by supporting two different conflict sides in Syria.

Imagine gameplay situation

You decided to conquer small weak nation with is behind in tech progress. You declared war and than you noticed that they have tanks that they should not have because other bigger AI that dont like you just gave them tanks when you declared war.

Imagine how cool and unpredictable gameplay could be with this implemented.

@Koshling
Can you teach AI to support another AI (or even human player) by giving units when:
Donor AI is not at war. Gifted AI has good relations with donor AI. Civ that declared war have bad relations with donor AI.
 
There is already one event that lets you give resources to a nation of the same faith as you who are fighting a war.

You can already go into the Diplomacy Screen and give units, although this is usually limited to what they are willing to accept which is the latest worker units.
 
@DH
I know but I talking about teaching AI to do that

With this feature gameplay will be much more interesting and it will force you to care about relations with other civs because if you not than you could be at unofficially war with everyone.
 
@DH
I know but I talking about teaching AI to do that

With this feature gameplay will be much more interesting and it will force you to care about relations with other civs because if you not than you could be at unofficially war with everyone.

Yes, it's a nice idea BUT.. first we'd need to do something about the current diplomacy system. It's very easy to get in bad relations with an AI (them begging war constantly) and very hard to improve them.
 
Yes, it's a nice idea BUT.. first we'd need to do something about the current diplomacy system. It's very easy to get in bad relations with an AI (them begging war constantly) and very hard to improve them.

It's not that hard to improve them, but you have to be prepared to take a number of very bad deals along the way in order to do so.
 
That kind of proxy war is really powerful. My brother once used it to great success against me in a vanilla game. He noticed me being at war with a fallen behind AI and gifted them all techs, so that they suddenly had a military on my level.
If unit gifting would have been possible I would have been toast.
 
Proxy Wars would help attack giant countries that seem unstoppable. It would be a good balance against this.
 
Do AI ships (talking about at or after Astronomy) ever take Maneuvring I? Without it you have halved movement across ocean. They take movement point increases, but an increase from 7 to 8 movement is no increase at all across ocean without Maneuvring.
 
The AI improvement is probebly the most importent thing for this mod.

i´m for one don´t feel realy threatend. I have only played up to emperor, not deity, but i still don´t think it´s going to change much. After getting better civics in the ancient era you can litterly explode. The AI dose not.
With like 30-50% or more of the worlds population. Having 100% on science and still making over a thousend gold a turn with 100´000 gold backup cash the game dosn´t really gice you a challange.

I play with the barbarian world and minor civ checked. And are also changing so the barbarians are as good as civs from the begining. This gives you a challange from the start but when you overcome that then you get to the point as just described.

Of course if the Ai would be better in staging war that would be a difference but i don´t think that is enough.
They have to be better at managing there cities, choosing right tecnologies and building more cities (when appropriate ofcourse).

For example: i´ve noticed that some AI´s build a tremendous amount of healers that just cost alot of money and dosen´t really do anything for them. They should only build them, with medic uppgrade, for keeping there cities healty.
Without getting too deep on this theme, because then this post is gonna be an essey :)
overall their cities should concentrate on hammers and building buildings that that give hammers first then buildings that are giving food. Ofcourse there have to be some tech building and units up in that mix. When they have some units and all the good building then they should "build science".
With tecnologies they should concentrate on getting tecnologies that really matters, like one religion, then tecs that give better civics.

Hope someone reads this, say what is good or piss on the things that they think is wrong with what i wrote. Just so that something is happening on this subject. Because you can change all the grafics or adding how many new features you whant. Without good AI the mod don´t give the player a challange. And that is sad for a mod that i think is AWESOME!
 
Do AI ships (talking about at or after Astronomy) ever take Maneuvring I? Without it you have halved movement across ocean. They take movement point increases, but an increase from 7 to 8 movement is no increase at all across ocean without Maneuvring.
I'm not sure... If you're certain they aren't putting enough emphasis on that promotion then I'd need to take a look at the tag it utilizes to see why and program something better. You're probably right and I can think of a few ways more emphasis may be placed on the tag appropriately. It'd just take some time to look into it.

The AI improvement is probably the most important thing for this mod.

i´m for one don´t feel realy threatend. I have only played up to emperor, not deity, but i still don´t think it´s going to change much. After getting better civics in the ancient era you can litterly explode. The AI dose not.
With like 30-50% or more of the worlds population. Having 100% on science and still making over a thousend gold a turn with 100´000 gold backup cash the game dosn´t really gice you a challange.

I play with the barbarian world and minor civ checked. And are also changing so the barbarians are as good as civs from the begining. This gives you a challange from the start but when you overcome that then you get to the point as just described.

Of course if the Ai would be better in staging war that would be a difference but i don´t think that is enough.
They have to be better at managing there cities, choosing right tecnologies and building more cities (when appropriate ofcourse).

For example: i´ve noticed that some AI´s build a tremendous amount of healers that just cost alot of money and dosen´t really do anything for them. They should only build them, with medic uppgrade, for keeping there cities healty.
Without getting too deep on this theme, because then this post is gonna be an essey :)
overall their cities should concentrate on hammers and building buildings that that give hammers first then buildings that are giving food. Ofcourse there have to be some tech building and units up in that mix. When they have some units and all the good building then they should "build science".
With tecnologies they should concentrate on getting tecnologies that really matters, like one religion, then tecs that give better civics.

Hope someone reads this, say what is good or piss on the things that they think is wrong with what i wrote. Just so that something is happening on this subject. Because you can change all the grafics or adding how many new features you whant. Without good AI the mod don´t give the player a challange. And that is sad for a mod that i think is AWESOME!

Surprisingly, the ai's tech pathing and building choices are quite good. They can be a little improved by continuous work on finding where they value or don't value particular elements enough and tweaking the structure but the design is, at it's basis, very intelligent. Where it falls down the worst right now is what units it chooses to build and how it chooses to use them... ugh.

Anyhow, we aren't ignoring these things at all... our AI programmer is on a small hiatus and knows he has a lot to do there - he's also one of our best all around programmers and is often wrapped up in debugging and optimization as priorities over the AI. AI work is not something anyone on the mod team can do really... I may be able to contribute more in that direction but for now I don't want to mess with things until I have some more of the restructured combat rules fully setup because those rules will influence differences in choices that should be made.

Point being, we're certainly aware of this issue and have plans to address it to the point where the ai will be a truly lethal opponent.
 
I'm not sure... If you're certain they aren't putting enough emphasis on that promotion then I'd need to take a look at the tag it utilizes to see why and program something better. You're probably right and I can think of a few ways more emphasis may be placed on the tag appropriately. It'd just take some time to look into it.



Surprisingly, the ai's tech pathing and building choices are quite good. They can be a little improved by continuous work on finding where they value or don't value particular elements enough and tweaking the structure but the design is, at it's basis, very intelligent. Where it falls down the worst right now is what units it chooses to build and how it chooses to use them... ugh.

Anyhow, we aren't ignoring these things at all... our AI programmer is on a small hiatus and knows he has a lot to do there - he's also one of our best all around programmers and is often wrapped up in debugging and optimization as priorities over the AI. AI work is not something anyone on the mod team can do really... I may be able to contribute more in that direction but for now I don't want to mess with things until I have some more of the restructured combat rules fully setup because those rules will influence differences in choices that should be made.

Point being, we're certainly aware of this issue and have plans to address it to the point where the ai will be a truly lethal opponent.

Okey sounds good :)

I have been playing realism invictus on the side abit. That mod don´t get as deep in every period in time with techs, units or building as C2C. So it is probebly easier to program the AI civs for that reason. Less options.
But with aggresivness and building up an army for an attak the Ai there make a pretty good effort I think.

On higher diff. level on RI the AI civs start is alot better then yours and that ofcourse gives them a great advantage and more of that is not what the goal should be.

Don´t know if some people working on C2C also have made work on RI, the people usely overlap somewhat. But perhaps there might be some ideas to be takend from there on army build up.
 
I don't know if it was mentioned here or elsewhere but since t is about AI I'll talk about it here.

The Immigrant unit has an automate function that identifies what city it should join and gives the orders to the unit to go there and join the city. This has the problem that the unit may move into danger. Koshling mentioned that this sort of activity was not used bu the ai anywhere else but checking has shown that two other classes of unit behave exactly the same these are the automated missionary and corporate executive both find a city that needs their attention, go to that city and perform their mission. All without regard to any danger along the way.

So is it possible for
1) the go to city part of these missions to be made aware of danger eg if the path crosses foreign or wild lands add the 'request escort" to the movement

2) call such a "go to" function from python so the Immigrant can use it.​
 
I'm playing a game on the GEM (Eons, Deity) and I'm about to reach the Industrial Era. I own the complete North and South American Continent, the Zulus have 4 glorious cities in Greenland and India has ~250 Cities in Europe, Asia, Africa, England, Australia, Japan... basically the whole rest of the world exept some islands in the pacific.
Right after the war with the zulu I had turn times from up to 25 minutes due the massive amount of troups they have. It's better now, around 8 minutes per turn, but still quite boring.
I have 1.4 times the troops of the Zulu (almost exclusive Guards, around 700) and India has a ratio of 0.1, so probably 10 times the troops I have! The Zulus are no thread anymore and I don't have any offensive unit at all. I was wondering if the AI is programmed that it calculates the troops and decides if it has already enough. If so, will it stop building troops or even delete some? This would strengthen their economy A LOT and if you see the situation logically, neither of us really needs to go to war anymore. We both have enough cities, we have plenty of all resources and trade a lot, we have a religion or even Atheism and there is no overlaying culture.

To avoid long turntimes in the later game it might be good if the AI tries to reduce their army if there is no thread and it determines the chance of war is very low.
 
The Congolese have had a Narwhal trapped in their culture under ice since at least the Ancient era. It is next to a roaded tile, and two tiles from a city. Archer Bombard is on, so a few archers could have killed it millennia ago. The AI either doesn't know, or doesn't think it's worth it.

I would kill it myself, but Archer Bombard doesn't work in friendly (as opposed to domestic) borders - even against enemy units - and the poor whale is already too injured for my Ornithopters to do anything (Ornithopters can bombard enemy units in friendly territory).
 
Here's something I've noticed in my most recent game during Ancient/Classical.

You can't attack a city until you've gotten a cities defenses down to a certain point. 60% bonus I think.

Problem for the AI (and me) is that the first siege unit, the battering ram, doesn't have a bombard option. It can only attack and hopefully knock a few points off the city defenses. Needless to say, this invariably proves fatal for the ram.

During the Ancient and Classical periods though, city defense scores are still capable of getting very high. You can need a very large number of rams, and you need to save them up to use on the same round else the war will regenerate, and the AI is not taking this into account when declaring war. I'm seeing wars all over the place with no cities changing hands because they just can't get the walls down.

Once you invent catapults, I expect this problem goes away, as they're re-usable.

Ideally, rams would have a "bombard" option that doesn't use them up. (I'm almost certain that they used to have this.) Either that, or the amount of possible defense bonuses available in the pre-catapult era might need toning down a little.

In a related thought, the AI has gotten very clever indeed about building and using forts. Forts cannot be bombarded, and do not care about your City Attack promotions, and as such, ones on hills are almost unassailable.

These two things combine to give defenders more of an upper hand than they probably ought to have
 
Are you finding that rams aren't able to knock down defenses much before dying even if they have developed out a number of promotions that assist them in doing so with added Breakdown Chances and Damages? Those could be set a bit too low as a base on Rams due to my initial tests showing it to be so strong.

I strongly disagree that it would be 'ideal' for them to have a ranged bombard. The whole concept behind this change is that a ram cannot damage city defenses without putting itself in harm's way.
 
Are you finding that rams aren't able to knock down defenses much before dying even if they have developed out a number of promotions that assist them in doing so with added Breakdown Chances and Damages? Those could be set a bit too low as a base on Rams due to my initial tests showing it to be so strong.

I strongly disagree that it would be 'ideal' for them to have a ranged bombard. The whole concept behind this change is that a ram cannot damage city defenses without putting itself in harm's way.

By "ideal", I mean for gameplay and balance, rather than historical accuracy. The promotions no doubt help, but as a suicide unit a ram's life is a short one, and they're not going to be earning any xp to gain more than they manage to start with.

It's not me I'm concerned for. I can just guess how many rams I'll need, build a big stack of them and march them all over together to use on the same turn. The AI doesn't seem to be wrapping its head around that though. Does it understand that rams are not normal siege weapons, and that it needs way more of them than it would catapults?
 
in my last game seem the comp is very willing to trade so many tradeable ships for a tech. i was ahead in science but wasn't building a lot of ships because couldn't get into ocean suddenly could and met other countries traded them a tech for a lot of their weak ships upgraded them and with 3 trades had decent sized navy.
 
Top Bottom