Late Game Conquest

[Thalassicus]

To put it another way, I want to make it easier for AIs to conquer AIs, without affecting human-AI combat.

The first and second parts are easy individually, but it's a struggle to combine them. The simplest solution is a strength bonus in AI-AI combat that depends on their "AI-boldness" ratio. A militaristic AI with 10 boldness would get a 100% strength boost against an AI with 0 boldness. This would take just a few minutes to do...

...if we had game core access.

Since we don't, I'm trying to think of another way to produce a similar effect.

 

[Txurce]

By that definition wouldn't buffing military buildings affect the player?

Only adversely. I'm proposing buffing those buildings for the AI only, in a similar way to how they now receive highly-promoted units (and we don't). I don't know if Thal can do something like this.

Truly the best option in my mind is to reduce the DoF RA bonus (I'm actually not clear why it was raised) or find some way for militaristic AIs to clique up like peaceful ones do (which has been hinted may happen more in G&K).

Doesn't reducing the DoF RA bonus hurt the peaceful human player? I can see how the net benefit would somewhat help militaristic AI, but not enough to counter the loss of the relative edge humans have in employing this tool (especially with the new, tougher VEM). In other words, I acknowledge your point, but wouldn't want to lose this!

 

[Txurce]

To put it another way... I want to make it easier for AIs to conquer AIs, without affecting human-AI combat. The first and second parts are easy on their own, but it's a struggle to find things that only affect AI-AI battles.

Would it be fair to say that adversely affecting human performance in warmongering would be a good thing? I'd think so, given that the AI is worse at fighting and conquering than anything else. That's why I'm proposing a further buff to their units, ideally only to those militaristic ones who build barracks, blast furnaces, etc.

 

[mitsho]

Only adversely. I'm proposing buffing those buildings for the AI only, in a similar way to how they now receive highly-promoted units (and we don't). I don't know if Thal can do something like this.

Dummy Building giving bonus to a class of buildings which is deleted for the Human Player?

Do AI warmongering player actually fill out the honor tree? If they tend to chose this tree automatically, one could just give them one or two free social policies (that they will chose in the honor tree). They are not aiming (mostly...) at a cultural victory, so it shouldn't affect that, right?

 

[Txurce]

Dummy Building giving bonus to a class of buildings which is deleted for the Human Player?

Yes, this sort of thing. Then the dial could be adjusted for just how powerful to make it. My gut says it should be pretty powerful, in order to synergize with the possible new Domination VC, and make these civs more than just a theoretical threat.

Do AI warmongering player actually fill out the honor tree? If they tend to chose this tree automatically, one could just give them one or two free social policies (that they will chose in the honor tree). They are not aiming (mostly...) at a cultural victory, so it shouldn't affect that, right?

This could also help, provided there were a way to keep the human from getting the same benefit.

 

[Thalassicus]

There's three types of confrontations:

1. Human vs militaristic
2. Human vs peaceful
3. Militaristic vs peaceful

I've buffed #1 and #2 in VEM as much as I think I can without making luck too big a factor. Starting near militaristic AIs is much harder than near peaceful AIs. It can frustrate people. What I'm trying to find now is how to improve situation #3.

 

[Txurce]

There's three types of confrontations:

1. Human vs militaristic
2. Human vs peaceful
3. Militaristic vs peaceful

I've buffed #1 and #2 in VEM as much as I think I can without making luck too big a factor. Starting near militaristic AIs is much harder than near peaceful AIs. It can frustrate people. What I'm trying to find now is how to improve situation #3.

That helps.

Keep in mind that militaristic AI do well enough at the start - they just start to fade as the game hits the 30-50% mark. So you could give them an era-based dummy building that buffs the military buildings. It could even be a series of dummy buildings, starting at the start of the Medieval era, and cranking up with each era.

 

[Thalassicus]

I think that would have the downside of making it harder for the human player to start warfare in the mid-late game. I want people to be able to attack AIs later even if they did not fight early wars.

 

[orangecape]

I think the simplest way to give warmongers a bonus is to improve the Honor tree in a way they can use. Players of course would take advantages of the same boosts, which isn't necessarily what you want, but maybe you could tie it into the military XP buildings? The AI builds those buildings in many cities and fairly randomly while a player will usually buy only 1 or 2 copies (AFAIK) for focused military cities or rushbuying. Adding "All Barracks, Armories and Military Academies give +1 Science and +1 Gold" somewhere in the Honor tree sure would help out AIs that build those buildings. This has a minimal effect for a human playing well since people aren't likely to build those buildings everywhere anyway but would be a much more substantial boost for the AI. Of course the bonus could be something more combat related but this would at least help militaristic AIs get ahead a little. I don't know if you can add "+15 XP from each Barracks, Armory, Military Academy" or not but that would accomplish the same aim (though the human player sure would take advantage of that!)

 

[Txurce]

I think that would have the downside of making it harder for the human player to start warfare in the mid-late game. I want people to be able to attack AIs later even if they did not fight early wars.

How much harder, given how relatively easy conquest is? As hard as competing in science is now? Probably not - and you could adjust the throttle if there were complaints.The benefit in AI vs AI warfare is obvious.

 

[Thalassicus]

Not conquest... starting warfare late.

Conquest is killing AIs from the classical era until we win. This is too easy, so your suggestion would help this by making it harder.

Starting warfare late is too hard, based on the feedback I've seen about runaway science AIs. Peaceful players should be able to fight a limited war to slow down runaways (pillaging and such) with a reasonably small military investment. Boosting AI-vs-human military performance would make this worse. This is why I want to focus on focus on AI-vs-AI changes, not AI-vs-human.

 

[Txurce]

Starting warfare late is too hard, based on the feedback I've seen about runaway science AIs. The suggestion would make this part of the game worse.

Do you mean by having to deal with tougher militaristic AI as well as runaway science AI's (who wouldn't benefit from the militaristic boosts)? If so, I see your point.

 

[Txurce]

The best option I can think of is lowering some of the bonuses that peaceful empires receive (such as lowering the DoF RA bonus).

I don't know if the WWGD change in v132 - which I interpret as nerfing the AI only, a good thing - is what Seek meant, but it's a welcome, no-downside change regardless.

 

[Thalassicus]

I think I figured out how to phrase what I was trying to say, and revised it.

 

[Txurce]

I think I figured out how to phrase what I was trying to say, and revised it.

I see what you're saying. Was it clear that my suggestion would only benefit militaristic also-rans, and not runaway pop/science civs?

 

[Thalassicus]

I'll try and explain it another way...

Say we're playing a game where the Isabella militaristic AI conquers her continent. With that many cities, she'll probably have more science and population than a peaceful human on the other continent. She also has a huge and highly-experienced army.

This is why we'll need to reduce AI-vs-human bonuses, not increase them. Otherwise it will be impossible for a peaceful human to win. The AI-vs-AI bonuses I'm searching for are different from the AI-vs-human bonuses suggested so far.

 

[mitsho]

Going back to squar one, the AI fails at conquest victories. If it takes a capital, which does happen, it takes long for them to mop the rest up and move on. Why? Some guesses, correct me if I'm wrong, I'm only a casual player :

• It loses too many units with conquest. Thus they value their own strength too low -> they are not self-confident enough to enter another war
• If they succeed in replacing the lost units, it's not good enough to bring them to the front fast enough. -> the last punch to take those last "mountain fortresses" () isn't there.
• It enters peace-treaties too easily and thus doesn't receive enough benefit for a 20 turn pause in warfare. -> It probably waits these turns out whereas the Human Player uses these for another border war with another AI or takes over a city state.
• If the warmongering AI gets to a good start (f.e. not getting his whole first army slaughtered by the Human Player), it then fails at building up its economic side to fuel the conquering machine more.
• The AI choses its targets mostly by strength, not proximity of its army. -> Long Napoleonic Trecks of Death deep into Russia instead of conquering that close civ/city state.
• Thal points out often that attacking a city state first is a good option for conquering games. Instead, the AI often suicides its first army against a Major Civ. The AI also doesn't have a Grand Strategy, a list of civs it wants to attack one by one, instead it often jumps depending on the situation at hand. When I play conquest, I often have a clear idea of my conquering plans.

I think, the biggest point may be that the AI fails at effectively snowballing: Finishing up one civ WHILE preparing for the take-over of the next one. Thus the golden opportunity of the unique units often are not used in full potential. So... (again, more debate questions that thought through remarks. I haven't played for two weeks or so)

1. Whenever an AI conquers a AI city/capital, it may receive a free unit?
2. The AI should (?) overestimate its own military strenght
3. the AI should not enter Peace Treaties with other AI easily. (I know, they never agree to a peace with ME either...), maybe this is a Leader specific thing?
4. Whenever a AI conquers a capital, it should receive a bonus (economic, scientific or cultural) that lets itself govern the new lands more easy thus, maybe similar to the Korean trait with science buildings?
5. The AI should value proximity stronger over strength with military decisions.
6. Design decision: Partisans are for turning the tide and helping AI defend itself better. With AI-to-AI-wars we want to support the attacking side though, right? That's kinda the opposite...

As for barracks and the unit cap, I would like that as it sounds realistic and it makes barracks more important. But, it might be not fun. At least I do hate it when I hit the unit limit... and I think it's a change that will not solve the problem at hand (or the benefit is too low for the cost of making the AI build them efficiently...).

 

[Thalassicus]

I realized the same thing you did about partisans, and included that in a number of small changes in v132.

#1 and #4 are possible, but I think the AI has a harder time starting conquest. It does well once it gets the ball rolling.

That reminds me why we've got the whole problem! About a year ago, Firaxis blocked AIs from attacking citystates in the early game. Even if I give Genghis Khan a dozen units near a citystate he's at war with, he won't go after it anymore. This hampered AIs in a way I've never been able to repair, despite trying for over a year. Remember back at the game's release when AIs would steamroll a whole continents? They can't get started anymore because they aren't allowed to pick on the citystates. I'm so used to it now I'd completely forgotten about that.

The ideal changes would be (if we had core access):

• AI combat bonus vs AI units.
• AIs target citystates first.

I've struggled for a year to find some AI-vs-AI alternative we can do. I'm out of ideas by now, so I'm hoping this discussion will strike on a possibility.

 

[Txurce]

With regard to #1 and #4, keep in mind that the capital-conquering AI is at least as likely to be a Persia as a Songhai. This is because most of the conquered capitals fall in the second round of wars, by which time the militaristic civs are already second-tier. You would then be boosting one of the pop/sci runaway AI.

 

[albie_123]

If other people like it too, I wouldn't mind a change in the domination victory - but that's not really a fix to this problem.

Could you be more specific on what 'Added a few changes to help AIs conquer other AIs (no changes to human-vs-AI warfare)' means in the patch notes?