Idea for improving AI's use of forts

[tlaurila]

Supposing you were correct however and it did make city conquest harder, would that be a good thing? It would limit the AI's ability to capture cities even more than the humans, because the human is better at bringing along overwhelming force, whereas the AI often brings just enough force to lose its army while not capturing the city.
[In particular here, I'm worried about the AI's ability to take cities from each other, not from the human. Weaker AI ability to take cities is not a wash, human-difficulty-wise, because it makes it harder for an AI superpower to emerge.]

This is a good point. We want AIs to be able to take cities so that some AI can become stronger and give opposition to the player in later eras. One way to have this have less effect on AI vs AI than human vs AI is to have the bonus be more against ranged than it is against melee, like I suggested. Humans use vastly more ranged fire than AI, at least in my experience.

But I would urge again; what is the problem you are trying to fix?
I worry that you're arguing for more defensive bonuses just for the sake of them.

It started as spin-off to forts discussion, but the "problem" would here be what you've been saying all along: It's too easy for the human to overrun the AI by tactical superiority and grow in strength. (thus must eg. have happiness to slow down that pace)

Better to slow down the pace of conquest militarily, making the human need to devote more units, and put those units more to risk as opposed to trusting bombards all the time. Making bombards less useful on offensive is a good way to make humans need to devote more resources to conquest (thus playing to AI strengths via economy cheats) and *gasp* even lose units at times!

 

[Ahriman]

If the problem is that human strategies that use lots of bombardment are too effective, then surely the right response is to weaken bombardment units (by lowering their ranged attack values).

I certainly think that the VEM buffs to crossbows were not necessary, and VEM longbows are totally insane, the best unit in the game by a long way. Longbows remain effective until the mid industrial (infantry/light infantry don't take much damage from longbows, but everything before then does).

 

[tlaurila]

If the problem is that human strategies that use lots of bombardment are too effective, then surely the right response is to weaken bombardment units (by lowering their ranged attack values).

But here you run into the catch that then AI's bombards will be equally nerfed. It's not strength of bombards itself, but the value of correctly positioning bombards to first clear a city of defending units by picking them off from outside the range of the city and it's ranged garrison, and then capturing the city. The picking usually involves catapults and horses (or advanced equivalents). This a human can do very, very well, and it means loss ratios of units between human/AI is what, probably less than 1/10.

Simply reducing ranged attack strengths, though I think some of this could well be done especially for the artillery line of units against other units and for crossbows, would likely reduce effectiveness of AI as well. And it would reduce the danger the AI presents to the human. Rarely do I lose a unit that isn't bombarded first, because knowing from where the AI can bombard come his turn is harder than melee attacks. Usually I lose units to a surprise bombard on a unit that I figured can withstand what threat it seems face.

 

[Ahriman]

But here you run into the catch that then AI's bombards will be equally nerfed.

But the AI is very ineffective in using bombardment, and uses them much less often. They'll use their siege units on my cities if I don't have an army, but I hardly ever have my units bombarded much, in part because the AI is horrible at screening and preserving ranged units, and in part because it builds fewer.

The only AI ranged attacks I ever get hit by a lot are those with 3 range.

In contrast to you, I normally lose units not to bombardment but to being unexpectedly surrounded; the bombardment doesn't do much of the damage, most of the damage comes from melee units. Many of the units I lose are also due to AI units with 3 or more movement or range (particualrly since it can be hard to tell which vanguards have 3 moves or not).

It's not strength of bombards itself, but the value of correctly positioning bombards to first clear a city of defending units by picking them off from outside the range of the city and it's ranged garrison, and then capturing the city. The picking usually involves catapults and horses (or advanced equivalents). This a human can do very, very well, and it means loss ratios of units between human/AI is what, probably less than 1/10.

Right, but weakening bombardment will make this kind of strategy less effective, and will make it harder to kill off AI swarms of melee units before they can take some of yours with them.

 

[Dunkah]

Adding a Fort/ Base to an area, can significantly increase the economy of that area due to the needs of the large host that is occupying it.

I like the idea of making Forts similiar to villiages/ trading posts. They should take a significant time longer to build than Villiages so that Ahriman's idea of spamming them early on will put a player behind improvement wise due to longer build times.

Also if it does not get the expansion bonuses from techs or policies that would make them less desirable later on. Perhaps a culture bonus later on as it becomes more of a tourist attraction.

 

[Ahriman]

<edit>
double post sorry

 

[Ahriman]

Adding a Fort/ Base to an area, can significantly increase the economy of that area due to the needs of the large host that is occupying it.

That doesn't make sense. Having to maintain a garrison is a fiscal drain, not an economic boon.

I like the idea of making Forts similiar to villiages/ trading posts. They should take a significant time longer to build than Villiages

I still think this kind of attempt is doomed. How is the AI supposed to know where a good place to build a fort is? Making this kind of decision intelligently relies on so many different factors; on terrain, on diplomacy, on relative military threats, etc.
The AI just isn't ever going to be able to combine all of these together, particularly given our very limited ability to influence the AI.

And if the AI can't place them intelligently, then it either won't build them, in which case what is the point, or it will build them haphazardly, in which case it is wasting worker time.

And if they have no upkeep cost, then the human can spam them later in the game (when they do have spare worker time) in every tile that they aren't planning to work.

 

[Dunkah]

That doesn't make sense. Having to maintain a garrison is a fiscal drain, not an economic boon..

Apparently you have never owned a store near a miliatary base. Try telling this to the communities that have lost the bases they relied on for a strong economy for years.

The garrison is represented by the unit inside it that cost maintenance all ready.

I still think this kind of attempt is doomed. How is the AI supposed to know where a good place to build a fort is? Making this kind of decision intelligently relies on so many different factors; on terrain, on diplomacy, on relative military threats, etc.
The AI just isn't ever going to be able to combine all of these together, particularly given our very limited ability to influence the AI.

And if the AI can't place them intelligently, then it either won't build them, in which case what is the point, or it will build them haphazardly, in which case it is wasting worker time.

Have the AI build them on unoccupied hexes on the borders of hostile Civs. (Or if they are Afraid). In front of hills being better than not. On a road bing better than not. On a point on a line from city to city. More often than not that would be decent. With a little thought could be better.

And if they have no upkeep cost, then the human can spam them later in the game (when they do have spare worker time) in every tile that they aren't planning to work.

Who said no upkeep?

 

[Ahriman]

Apparently you have never owned a store near a miliatary base. Try telling this to the communities that have lost the bases they relied on for a strong economy for years.

You are mistaking local economic benefits from aggregate economic benefits.
If the central government pays a whole bunch of people in area X, then that is good for people in X, but it comes at the cost of all the other taxpayers. This is a transfer payment.

The garrison is represented by the unit inside it that cost maintenance all ready.

So, you're saying that a military unit provides economic benefits if it is in a fort, but not if it is in a city, or in a village or farm or mine? That doesn't make sense to me.

Have the AI build them on unoccupied hexes on the borders of hostile Civs

How? We don't have access to code that could do that. Worker task code isn't connected to diplomacy code.
Worker code also isn't connected to terrain preferences.
People have tried getting AIs to use forts intelligently before - in Civ4 for example, where we had lots of code access. It is very hard to do.

Who said no upkeep?

It doesn't really make sense to have both upkeep and gold income. They cancel out.
And if they have upkeep, we're basically back to the status quo, where they aren't worth building, especially if they aren't near a city.

 

[Thalassicus]

- Can we improve forts so that they are worthwhile, bringing a stale element back into the game?
- Can we make the AI use forts somehow usefully?

Any significant changes to forts would require the game core only Firaxis has access to. The discussion is just hypothetical... for whenever we do get core access.

I certainly think that the VEM buffs to crossbows were not necessary

VEM actually nerfs crossbows -2 compared to vanilla. I think I reduced them -1 in July, then another -1 in August. I can take this further if you feel it's needed.

 

[Ahriman]

VEM actually nerfs crossbows -2 compared to vanilla.

Really? Did they get buffed in vanilla then? I didn't follow the vanilla changes closely enough.
Because in release vanilla the longbow and CKN had ranged strength ~10, and then 12 on crossbows.

 

[Thalassicus]

In... June I believe, whenever the big Firaxis "balance patch" came out, they removed the -25% damage penalty ranged attacks used to have. They also added a large strength bonus to all crossbow class units (xbow, ckn, longbow). I felt the double buff was excessive and have been nerfing it since then.

I think crossbows are currently at 13 in VEM, longbows at 12, and ckn 9. I could drop them another -1... I've been doing small increments because I don't want non-UU crossbows to go back into the "useless" territory of vanilla release.

 

[Ahriman]

I think crossbows are currently at 13 in VEM, longbows at 12, and ckn 9.

Really?
I could have sworn that crossbows and longbows were at ranged strength 14. [I haven't played China for a while.]
I'll check tonight.

I didn't find them useless in vanilla.

 

[Thalassicus]

It's only regular crossbows that were a problem, but that's much less of an issue now that Machinery has multiple useful things on it. The lumbermill boost, foundry, and armory are post-release.

Looking at the files I see...

Vanilla:
15 Crossbows
15 Longbows
11 CKN

Vanilla Enhanced:
-2 for all crossbow class units
-1 extra for Longbows

I haven't played China or England in a while, so I'll double check this ingame to see if I'm overlooking something in the files.

Update: I accidentally reduced the combat strength of these units, not ranged combat. Oops! I've fixed this bug for v108.1.

 

[Ahriman]

Update: I accidentally reduced the combat strength of these units, not ranged combat.

Ah! So were both sorta right and sorta wrong.
Cool. I think that will be good for balance, these guys are so deadly. I have wiped out entire civs with only 4-5 longbows and a levy.

 

[tlaurila]

I still think this kind of attempt is doomed. How is the AI supposed to know where a good place to build a fort is? Making this kind of decision intelligently relies on so many different factors; on terrain, on diplomacy, on relative military threats, etc.
The AI just isn't ever going to be able to combine all of these together, particularly given our very limited ability to influence the AI.

I think you're shooting at the wrong target here: Do not ask yourself how is the AI going to play like a human. It just can't. Ask what does the AI need to do improve the game it is currently playing. AI can't put down forts smartly like human, but it doesn't mean it's current game couldn't be improved by not so smart fort placement.

If the AI would, dumbly, just build a couple forts next to every one of its cities, and seek to maintain a garrison in these 3 tiles, I think it would improve the AI's ability to withstand extermination by the human. Without it being so easy to pick off all units around the city before assaulting it, the human would much more likely face losses in the attack as the AI counterattacks.

If the problem is that AI can't create large enough forces to attack and take cities for the love of anything, then surely the answer is to edit the AI to muster more forces before attacking. Not keeping city defenses so low that the human can coast over a continent without a sweat.

Agreed that a fort can't simply be a village with a defensive bonus and longer construction time. Spamming them would be silly. But if they would give 1 gold instead of costing 1 (still having a cost relative to village), and couldn't be constructed to neighboring hexes, I think it would be an improvement (pun intended ). It's not at all unreasonable that the protection offered by a fortification increases trade in early eras. And in later eras, the civilian infrastructure that grows around a garrison is capable of wealth creation as well, not just soaking up trickle-downs from tax money spent on the soldiers.

 

[Ahriman]

I think you're shooting at the wrong target here: Do not ask yourself how is the AI going to play like a human. It just can't. Ask what does the AI need to do improve the game it is currently playing. AI can't put down forts smartly like human, but it doesn't mean it's current game couldn't be improved by not so smart fort placement

I think you're shooting at the wrong target. A change that helps the AI isn't really a boost to the AI if it also boosts the human player just as much or more.

If the AI would, dumbly, just build a couple forts next to every one of its cities, and seek to maintain a garrison in these 3 tiles, I think it would improve the AI's ability to withstand extermination by the human

But it would weaken its economy, so it would have (slightly) smaller cities, fewer units, less tech, fewer buildings, fewer policies, etc.
If the fort has a maintenance cost, the effect is even worse.

If the problem is that AI can't create large enough forces to attack and take cities for the love of anything, then surely the answer is to edit the AI to muster more forces before attacking.

No, not really, because congestion problems kick in, and adding more force really doesn't make much difference. Sending a huge swarm of units into an artillery killzone where they die every turn isn't going to be changed by adding a few more units.

But if they would give 1 gold instead of costing 1 (still having a cost relative to village), and couldn't be constructed to neighboring hexes

Without a maintenance cost, I can spam them freely in every tile that I am not planning on working, giving a huge advantage to the human player on defense.

 

[tlaurila]

But it would weaken its economy, so it would have (slightly) smaller cities, fewer units, less tech, fewer buildings, fewer policies, etc.
If the fort has a maintenance cost, the effect is even worse.

This is extremely easy to answer: More economy bonuses and/or harder difficulty! If we can improve AI tactical ability at cost of these, it should be a no-brainer to go for it.

 

[Ahriman]

So the argument is... the AI is too weak, so lets add something which weakens it relative to the human player, and then move to a higher difficulty level with higher bonuses.

Why not just skip the middle step?

I don't see why having lots of fort improvements dotting the landscape is a desirable goal.

 

[tlaurila]

Why not just skip the middle step?

Oh, come on, you know the answer to this. To quote yourself:

No, not really, because congestion problems kick in, and adding more force really doesn't make much difference.

AI tactical capability and it's economic capacity are widely different beasts. You know this. And if we can give it some tactics at the expense of economy, compensating for the latter is trivial. Tactics is also in-your-face to the player, whereas AI economic bonuses are "under the hood".