AIs and the Art of War

Dancing Hoskuld said:

There does need to be some changes to the AI on the hunter line anyway. They seem to be a turn behind in their decisions, they continue to move when there are perfectly good and even wounded animals right next to them. They prefer to move into a good defensive position and let the animal attack, which often it wont. They don't avoid danger.

After Canine Domestication the hunter units should be considered to have a dog present so they should see invisible. This should reduce their loss to ambushers. (I loose lots to ambushers when they are on auto hunt ).

Click to expand...

1)The hunter is using basic fundamental AI functions for those kinds of decisions. For the most part it looks like it evaluates things in a proper order but something isn't quite right about the values it calculates out for various determinations. That being the case, this would be affecting nearly all the AIs so if it were sorted out to be greatly improved for hunters it would probably improve things a lot for other AIs too. Easier said than done of course. Situations need to be caught and repetitively evaluated so it ends up being a pretty solid project for even a small tweak. Yes, we'll need to address these things... I'm just saying it's going to take some major patience and very slow development.


2)At that point who needs dog units? I'd prefer it if canines could join hunter units as escorts... like Lord units can attach to any other types, adding their ability to see camo units (and some pusuit and strength) through the attached promotion that comes from this sort of union.

Also... besides killing hunters, are ambushers worth much for anything but cleanup after a major conflict weakens a stack? Ambushers SHOULD be the big problem for hunters shouldn't they? At least with a recon escort they will have a buffer against these attacks. Might be a good thing to also request a canine escort provided the canine is fast enough to keep up.

I've been thinking that we need a new unit AI for spotter units like canines - and for naval versions particularly. It would help tremendously for helping such units focus their promos in more beneficial ways and would also help for properly distributing and positioning units that fill this role. IMO basically any role that selects its own promotions and behaviors differently and needs to be included in a measured degree with the efforts of other roles (in the case of Canines, hunting, city attack and city defense as well as potentially a patrol role) should be defined as its own AI type.

I had thought of a canine and hawk unit being able to attach to a hunter but could not figure out how to limit it to hunters and perhaps recon units.

Which reminds me the animal healer is next to useless for animal units, especially mounted because they can't keep up with them.

Dancing Hoskuld said:

I had thought of a canine and hawk unit being able to attach to a hunter but could not figure out how to limit it to hunters and perhaps recon units.

Which reminds me the animal healer is next to useless for animal units, especially mounted because they can't keep up with them.

Click to expand...

Only the first versions of the animal healer can't keep up. Eventually they become mounted units that can. So at first they're best for keeping in the city or as part of larger attack stacks that also include the animals.

I wonder if restricting the promotion that comes from attachment to those types of units would do the trick? I could look into the code on the matter but it might be easier to just test that first.

Thunderbrd said:

I wonder if restricting the promotion that comes from attachment to those types of units would do the trick? I could look into the code on the matter but it might be easier to just test that first.

Click to expand...

Of course if that works we could get rid of all the different animal workers and just allow you to upgrade them with animal units.

That would be cool

Dancing Hoskuld said:

Of course if that works we could get rid of all the different animal workers and just allow you to upgrade them with animal units.

Click to expand...

Just watch out you don't lose too much game flavor in simplifying things. Unit stats are the foundation of a game but diversity and details create historical immersion. A good history game has both.

Dancing Hoskuld said:

Of course if that works we could get rid of all the different animal workers and just allow you to upgrade them with animal units.

Click to expand...

That may be a very good idea DH.

Why? Because even now I don't use all the animal workers. Human to Dog to Elephant and it takes quite awhile in game to upgrade/replace my human workers to Dog or Elephant workers. And Dogs are getting used less now too, because trying to upgrade a Human worker force is too expensive. And to remove that worker force and replace with better can be too time consuming. But is usually what happens.

JosEPh

I tend to keep my workers in groups of several instead and in Size Matters games merge them so each merged group is better than an animal worker anyway. So like Joe I seldom to never upgrade my workers (cost is too much for the minimal gain) but once animal workers are available I build of those when I need new ones.

Speaking of Size Matters; how will that work? 1 Tamed/subdued animal per worker group or one per base worker in a group?
If there is any way to make it happen the amount of animals needed should be 3x the group size, and the promotion should not be transferable when merging other workers with an animal promoted worker, aka still making them into a new unit or I could use one Tamed/subdued animal to improve n amount of workers. Micromanagement needed but if having 27 worker units (9 or 3 groups in Size Matters) building 1 Buffalo or 27 Buffaloes is an awful lot of hammers difference.

Cheers

BlueGenie said:

I tend to keep my workers in groups of several instead and in Size Matters games merge them so each merged group is better than an animal worker anyway. So like Joe I seldom to never upgrade my workers (cost is too much for the minimal gain) but once animal workers are available I build of those when I need new ones.

Speaking of Size Matters; how will that work? 1 Tamed/subdued animal per worker group or one per base worker in a group?
If there is any way to make it happen the amount of animals needed should be 3x the group size, and the promotion should not be transferable when merging other workers with an animal promoted worker, aka still making them into a new unit or I could use one Tamed/subdued animal to improve n amount of workers. Micromanagement needed but if having 27 worker units (9 or 3 groups in Size Matters) building 1 Buffalo or 27 Buffaloes is an awful lot of hammers difference.

Cheers

Click to expand...

Anything that causes a unit to sacrifice itself requires the unit be at its normal group volume as if it had just been trained to be able to do so. Further micromanagement should be unnecessary. It might be manipulable somewhat but the benefits would not be quite equivalent to just adding the sacrificing unit's strength to the group. Maybe a +1 or so small strength bonus (though with workers not yet) and some work speed bonus and the benefit of an added unitcombat so the workers can start taking some promos that stem from the new animal's unitcombat should be sufficient.

During my last campaign at deity i noticed that kept sending in strength 6 infantry and horsemen(i think they where called bandit raiders) into my lands near cities filled with strength 12+ cavalry units and a great general leading to my army getting stronger without hurting me at all.

Maybe this save will be useful in showing a problem. The Japanese are at war with me and they have a huge army outside one of my cities. However this army is almost all bombards (like 430 of them!). They will attack next turn with their other troops and I will win, leaving their 430 bombards to be killed, very slowly, by me.

This is with a very recent SVN. I know that Thunderbird said he had some work on this, but this SVN is from after then.

T C said:

Maybe this save will be useful in showing a problem. The Japanese are at war with me and they have a huge army outside one of my cities. However this army is almost all bombards (like 430 of them!). They will attack next turn with their other troops and I will win, leaving their 430 bombards to be killed, very slowly, by me.

This is with a very recent SVN. I know that Thunderbird said he had some work on this, but this SVN is from after then.

Click to expand...

The question is when this army would've been built relative to the updates being made. That update was very recent so would this whole force have been built with the update in mind? It is certainly possible that the AI would still be overvaluing such units though. The change I made was a light one that didn't go deep into the code where previous changes to get a higher valuation on these units had been hard coded and may need to be pulled back on some still.

I think it was recent. The Japanese had attacked me with similar armies previously with trebuchets and siege wagons, but I had mostly destroyed those. They only developed bombards recently.

Another issue. If you continue to play, you will see that the Japanese just leave their army by my city, with no chance of taking it, and just slowly being destroyed. I think it would be better if the AI could withdraw in these situations to protect itself.

I recently discovered another miracle feature that has potential to cripple the AI players, the iBuildUnitProb tag for the Leaders.
Maybe it is nice to have a little diversity but it's values range from 15 to 50 that is just to much. Since it has a influence in the production choice, all the leaders with low iBuildUnitProb can only suck in things like war, hunting, defense and everything else that depends on units. Leaders with higher values could just build too much units but that seems to be an advantage so it doesn't really hurt them.

I would just remove this tag or at least give all leaders the same value.

Early game AI:
playing on emperor (not agressive AI) the AI seems not to expand enough. The best AI reached sedentary lifestyle shortly before me (made a mistake by postponing its research a bit too long to keep my wonders alife)
Even that I has only 1 city in addition to the capital. (I have 4 + capital)
AI is apparently making too many military units there are some stacks in his territory but I did not find a settler. I would assume that by not putting aggressive AI it should have more focus on civilian builds AND settlers.

Is this AI issue already known?

I find it also surprising that the tech of the AI is not much behind due to so few cities. (Tech diffusion is turned on but I have no idea how much it does)

ichbinsehselber said:

Early game AI:
playing on emperor (not agressive AI) the AI seems not to expand enough. The best AI reached sedentary lifestyle shortly before me (made a mistake by postponing its research a bit too long to keep my wonders alife)
Even that I has only 1 city in addition to the capital. (I have 4 + capital)
AI is apparently making too many military units there are some stacks in his territory but I did not find a settler. I would assume that by not putting aggressive AI it should have more focus on civilian builds AND settlers.

Is this AI issue already known?

I find it also surprising that the tech of the AI is not much behind due to so few cities. (Tech diffusion is turned on but I have no idea how much it does)

Click to expand...

It is known but people's experiences seem to vary. Settings seem to play a role.

I found removing barbarians helped a lot for some civs

ichbinsehselber said:

Early game AI:
playing on emperor (not agressive AI) the AI seems not to expand enough. The best AI reached sedentary lifestyle shortly before me (made a mistake by postponing its research a bit too long to keep my wonders alife)
Even that I has only 1 city in addition to the capital. (I have 4 + capital)
AI is apparently making too many military units there are some stacks in his territory but I did not find a settler. I would assume that by not putting aggressive AI it should have more focus on civilian builds AND settlers.

Is this AI issue already known?

I find it also surprising that the tech of the AI is not much behind due to so few cities. (Tech diffusion is turned on but I have no idea how much it does)

Click to expand...

Is this coming from a current SVN or v35?

v35 (I downloaded C2C from the link in the main modpacks thread)
from your comment it seems that you are working on the ai? nice...

ichbinsehselber said:

v35 (I downloaded C2C from the link in the main modpacks thread)
from your comment it seems that you are working on the ai? nice...

Click to expand...

Yeah. I believe that part of the AI has gotten a lot of focus and SHOULD be working a lot better at this point.

Just walked through 3 Babylonian villages with only a couple of defenders each. Used two elephant riders to take out the entire civ-

frustrating thing no 1 is they had plenty of elephants, just hadn't developed the resources.

frustrating thing no 2 is that they had no Walls or other defenses and inappropriate units. Their Capital had a Tribal Guardian and an Axeman