Statecraft too popular

[JamesNinelives]

Ah, you mean complaints about certain policies being picked by the AI. That makes more sense. I had the impression you were talking about complaints in general.

I haven't hear it all that much recently, but yeah it does come up somewhat regularly.

 

[azum4roll]

There are 6 Statecraft and 2 Fealty this game again, even when Venice is present (and performing extremely well).

Unrelated, but 2 of the 5 religions got completely wiped out, one of the holy cities still held by the original founder.

 

[Stalker0]

I'm not super active any more but I was for like a year and a half, and not a single patch went by without someone complaining about some policy or ideology being picked by all the AI, and I can't remember a single case where it was an actual bug. I know it's anecdotal, but I would bet money that if I searched through patchnotes and other threads I could find well over 300 complaints and I doubt I'd find a single one with logs proving any sort of bias or overvaluing. If anything I think the real number is 100%, but I figured I might have missed a single patch where some sort of bug existed that actually caused the AI to mistakenly overvalue a choice.

I agree. Ultimately something like this is just much easier to test with bulk AI runs, and will always be more accurate than human reports.

 

[tothePAIN]

I agree. Ultimately something like this is just much easier to test with bulk AI runs, and will always be more accurate than human reports.

To some extent.

On Deity, I can predict with a high degree of certainty the policy choices that most civilizations are going to pick. Statecraft is the dominant pick. Games where the Fealty / Artistry are represented by 2 or 3 AIs are very rare. There are some AIs that are agnostic about their 2nd policy tree and do have some variation, but otherwise, its pretty consistent, and statecraft is high right now.

I don't notice this issue at Emperor, where it seemed like the AI was more likely to pick Artistry / Fealty. But on Deity, it definitely feels / seems like an issue.

 

[azum4roll]

Maybe it's map-related. Since Statecraft has a policy that gives strategic resources, AIs might give it more weight if the map lacks them.

 

[JamesNinelives]

Maybe it's map-related. Since Statecraft has a policy that gives strategic resources, AIs might give it more weight if the map lacks them.

I was thinking it might relate to the number city-states on the map. I see it taken in my games quite a bit, but IMO that's a logical decision when there are many city-states.

 

[DeAnno]

I think it isn't just the number, but the number *met*. I play on Oval and by 2nd tree most AIs have met almost all the CSes, so that could be why it's very popular consistently for me.

 

[Jean mi]

From my experience, AIs on "plus" maps where City States are separated on islands pick Statecraft a lot less often, I guess they decide to pick Statecraft when they have kind of easy access to City States.

 

[JamesNinelives]

From my experience, AIs on "plus" maps where City States are separated on islands pick Statecraft a lot less often, I guess they decide to pick Statecraft when they have kind of easy access to City States.

That's interesting. I play those maps and Statecraft still seems popular in my games. Game I'm playing at the moment on Continents Plus has 12 players - 7 Statecraft, 2 Artistry, 2 Fealty and me (Artistry).

 

[Stalker0]

I think I can sum this thread up:

@Gazebo if you could do us a favor. Next time you do a large series of AI runs could you check if statecraft is getting picked too often.

 

[Gazebo]

I think I can sum this thread up:

@Gazebo if you could do us a favor. Next time you do a large series of AI runs could you check if statecraft is getting picked too often.

I've adjusted it a bit for the next release.

G

 

[azum4roll]

I think I can sum this thread up:

@Gazebo if you could do us a favor. Next time you do a large series of AI runs could you check if statecraft is getting picked too often.

The post-turn 500 yield bug is hurting my interests to do another of those (or finish my current one) since King AIs can't win before that most of the time.

 

[Padre19]

I've adjusted it a bit for the next release.

G

When ?

 

[InkAxis]

When ?

Be patient man. You asking won't make it come out sooner, and remember everyone works on this for fun.

But anyway I agree that the AI statecraft pick should be slightly adjusted, but we definitely need AI runs with actual numbers before we make any major changes.

 

[azum4roll]

There are two possibilities:

1. AI rates Statecraft too high. Solution is tone down AI values.
2. Statecraft is too strong. Nerf Statecraft accordingly.

 

[InkAxis]

There are two possibilities:

1. AI rates Statecraft too high. Solution is tone down AI values.
2. Statecraft is too strong. Nerf Statecraft accordingly.

If statecraft is too strong, that's to be discussed elsewhere, in a thread about policy balance. This thread is about how often the AI chooses statecraft.

 

[azum4roll]

If statecraft is too strong, that's to be discussed elsewhere, in a thread about policy balance. This thread is about how often the AI chooses statecraft.

If AI chooses Statecraft because it's the strongest tree, it still belongs here.

 

[InkAxis]

If AI chooses Statecraft because it's the strongest tree, it still belongs here.

The AI doesn't analyze the strength of the tree before picking it. No AI on earth is that smart. If you mean what influences the AI to pick it then yes, that belongs here.

 

[ElliotS]

The AI doesn't analyze the strength of the tree before picking it. No AI on earth is that smart. If you mean what influences the AI to pick it then yes, that belongs here.

AFAIK The AI does in fact look at what yields a tree will grant, take into account their current plans vs the tree's properties, and choose based on that information. (With lower difficulties being more random, and Deity always picking the best option.)

 

[InkAxis]

AFAIK The AI does in fact look at what yields a tree will grant, take into account their current plans vs the tree's properties, and choose based on that information. (With lower difficulties being more random, and Deity always picking the best option.)

Yeah, I believe the AI looks at it's situation to pick trees, ex. Tradition when there is little land, statecraft when there are lots of CS, which can definitely be tuned (how many CS is "a lot"?). Can someone on the dev team confirm that the AI analyzes yields when picking trees? I feel like that would be difficult to code, especially as a lot of the bonuses are based on circumstance, and some don't even give yeilds, like giving spies for example. But if it really is like that, that's really impressive.