Belief selection & prioritization by the AI.

[SuperNoobCamper]

I know confirmation bias is a thing but what's with the first AI to found a religion ALWAYS choosing Orders as the first follower belief no matter what civilization, policy tree or the preferred win condition of that civ ?
In my last 5 games [2 as Japan, one as Indonesia, two as Spain] i did not found a religion first in either of them, always second because another AI gets a religion like 15 turns earlier than anyone else and in the last 5 games the 1st follower belief was always Order chosen by India [Tradition policy], The Celts [Tradition policy], The Huns [Authority], Unmet civ probably Songhai founding Islam with god of war, The Maya [Tradition].
I don't want to be jumping to conclusions so i'd really appreciate if you could share the results from your games where an AI founds a religion first in the following format to understand it better ?
First AI to found a religion :
Ancient policy tree :
Founder belief :
First Follower belief :
Diplomatic relationship with other civs if you are using info addict :

 

[Delvemor]

This is only anecdotal evidence, but even if I prioritize religion and faith generation, I've never been able to pick Orders when founding a religion, either on Emperor or Immortal, after a few dozens games on VP. Synagogues are always available though, and I tend to pick them by default and then get out of my way to maintain WLTKD. I never paid too much attention to which AI pick Orders but I remember a game as Mongolia in which I conquered indian cities with Orders, even though Ghandi was peaceful as usual, and he had the Tradition opener too. I ended up adopting his religion.

 

[CrazyG]

The AI does LOVE orders and they really aren't the best choice unless you are aggressive. Orders are very dependent on having a faith sink, like zealotry, that most civs by default do not have. What does India spend that faith on exactly?

 

[pineappledan]

I have been repeatedly playing the first 100 turns with my Phoenician civ. 6 games today. The AI players ALWAYS pick God of War first. It is never missed.

 

[Cokolwiek]

Synagogues are always available though, and I tend to pick them by default and then get out of my way to maintain WLTKD.

It is a proof that every civilization made by Firaxis is anti-Semitic.

Oh vey, it is like in the real world!

 

[Randomized]

I made a post about this issue (the orders, not the anti-semitism) a while back. It's definitely not just confirmation bias. It's pretty frustrating because I always play warmongers and I'm always desperately rushing religion so I can get orders, but someone founds a few turns before me and snatches it.

However, a few people argued that this was optimal play from the AI because they denied orders to a warmonger, who would have been very scary with them, and increased their own defensive capability at the same time. And I can definitely see where this argument is coming from. Even if it is rather annoying, isn't one of our big goals here to have an AI that is challenging because it makes good decisions, rather than because it has a huge amount of bonuses?

Now I doubt the AI is evaluating whether they are likely to be threatened by a warmonger and basing their decision on this, but even so there is likely to be one major warmonger in every game so denying them orders will always be somewhat relevant.

Anyway, this all comes down to whether the value gained from denying orders to someone else is greater than the value lost by not picking a more suitable belief. Unfortunately I don't know the answer to this.

 

[SuperNoobCamper]

Started another game, Ghandi again founding 20 turns earlier than me, guess what first follower did he choose this time ? Orders.

 

[crdvis16]

I think the AI should play the way a human would. I don't think most players would pick orders just to deny it to a warmonger. Players tend to pick beliefs for synergy and in support of their overall plan.

There should probably be some general rules for AI belief selection. Things like:

1. Don't pick orders (or god of war or hero worship or crusader spirit...) unless you plan to war a lot (chose authority or have bonuses in your kit for war or your leader personality is above average for aggression).

2. If you are India, Celts, or Spain then don't pick any beliefs around missionary/active spreading (India) or around inquisitors/having followers of other religions (Celts and Spain).

There's probably a bunch of other rules that could cut out some strange choices by the AI. I'm sure one can think of justifications for odd choices (like denying orders to a warmonger) but even with that reasoning I think it's better to just have the AI act more like the player.

In my current game the Netherlands picked Hero Worship. Sure- he has a decent UU but in general he isn't going on some conquering spree. His kit in general favors peace and trade. I'm playing as Denmark and I'm his neighbor... thanks for the Hero Worship holy city!

 

[CrazyG]

This whole "take orders to deny a warmonger" really isn't a very good argument. What you actually did was deny yourself (as an example) 1 science per two followers.

Sure you can find a scenario where an odd combination of beliefs was correct. I've taken God of War as tradition before, and I think it was a good choice. But that was an odd circumstance, if you do that every single game you're an idiot. 99% of the time mastery, cooperation, or creativity are going to help that AI Ghandi a lot more. Beliefs like scholarship are just good for anyone basically anytime.

Also, in my experience, he probably isn't denying it to a warmonger. He is denying it to the civ with the 2nd religion, which is usually like Siam or Ethiopia with tradition.

 

[SuperNoobCamper]

In my experience even pantheons seem to have some sort of prioritisation by some civs; Songhai never picks anything but god of war and seem to always get it first, Korea & Bablyon always go for godess of wisdom.
I like how these specific cases seem logical as it actuallu synergizes with their kit and game plan.

I have been repeatedly playing the first 100 turns with my Phoenician civ. 6 games today. The AI players ALWAYS pick God of War first. It is never missed.

2. If you are India, Celts, or Spain then don't pick any beliefs around missionary/active spreading (India) or around inquisitors/having followers of other religions (Celts and Spain).

Despite always founding first, India seems to always choose the Reliquary wonder tho they don't have access to missionaries and spreading via prophets even for india is expensive and counter productive, they would get more culture and tourism from building holy sites.

 

[Kim Dong Un]

My last game had me barely founding on emperor with Goddess of Protection, and surprisingly Orders were still available. I then proceeded to conquer the world

I agree that most of the time it's not ideal selections on the AI's part.

 

[LifeOfBrian]

I wish some things re: AI's belief selection was hard-coded. Like India not picking anything with missionary/GP spread to other civs or Spain&Celts not picking Pagodas or anything else based on foreign religions in your cities.

 

[Recursive]

I did not argue that it was optimal play, just pointing it out as a potential upside. I doubt that's the reason the AI chooses it.

 

[Gidoza]

Not exactly relevant per se as I founded first in this past game - I am a Warmonger and I did not choose Orders when I could have because I needed the Farms and Pasture bonus more. Someone else took Orders then even if I would have been happy with it as a second choice, but I opted for Mosques second.

The reason I mention this is because I don't feel like I really lost anything. Every single military unit I have ever trained has come from my Capital, do the Heroic Epic is giving me the Morale bonus anyways. I might have less overall Faith from not having Orders, but the other bonuses are more than making up for it and the 15 XP is not that big of a deal. The heart of it is that with Zealotry, my Capital can always train everything, so a bulk of the bonus of Orders is irrelevant. So the argument for "denying a Warmonger" is plainly a bad argument.

 

[JamesNinelives]

In the game I just played, the Celts were the first to found, they took Progress, their founder was Way of the Pilgrim and their first belief was Orders.
The second to found were the Maya, they also took Progress, founded with Ceremonial Burial and took Cooperation.

I was a little annoyed as I had hoping to found first; I built my second city next to Mt. Sinai. I got somewhat unlucky otherwise though, and didn't produce nearly as much faith as I had hoped. Founded sixth in the end after Carthage, Arabia, and Poland.

 

[BiteInTheMark]

I can confirm that orders are very often taken, coincidentally I noticed in my last game it was taken by a peaceful tradition civ as well some games earlier by India.
India in the same game picked a missionary spread founder and celts picked syncretism enhancer.

Can't say much about God of War, cause Iam rarely a warmonger. But I noticed, that God of commerce and God of wisdom are taken nearly every game and in most cases very early. Atleast for me indicating, that those are too strong, and it's my opinion too.

I think, in the best case the AI know which beliefs are bad for them. But I also think, this can't be soft coded, so a hard code should be the only solution and in my case, don't have any problem with it, as long as this is hard code is communicated to others which mod or change beliefs afterwards.

 

[Gazebo]

I fixed a bug that was introduced in December that was causing buildings and XP to be more valuable than they should be for belief selection. Should help with this a bit. It is hard to diagnose because there are literally hundreds of potential attributes that have to be considered for each building, and hundreds of attributes for beliefs/policies. Since there’s no unified method for accounting for these you have to generate unique data for each and every type of object being weighed by the AI. Furthermore you have to balance future planning with current needs, which is not always easy. There’s no ‘one cut’ solution here, and @BiteInTheMark if you want to understand that you can look at the code.

G

 

[Cokolwiek]

But I noticed, that God of commerce and God of wisdom are taken nearly every game and in most cases very early. Atleast for me indicating, that those are too strong, and it's my opinion too.

They are not, in fact they are perfectly balanced. They are garbage beliefs in the beggining, they give next to nothing both for wide and tall playstyles, but turtle to become mediocre in medieval, really nice by renaissance, and outright excellent by industrial. Sometimes synergy with scholarship, mosques, and jesuit education can proppel them to amazingness even faster, and this is for what you should strive in every religion and can basically do with many beliefs depending on your needs. It is a matter of planning and prioritizing. They are far from being too strong, I consider fertility, beauty, and war pantheons, sometimes even carnival and nature ones, to be better.

Why should you expect AI to not to choose strong and optimal route for them? You have the same oportunity to grab those pantheons and sculpt strong synergetic religion even on standard deity. If you want to have strong religion focus on it, don't whine that it is rewarding for those who sacrifice other things to prioritize it.

 

[JamesNinelives]

Not sure if it's still relevant, but here's some info from another game.

The Aztecs founded first with Scholarship and Way of the Pilgrim, having taken Authority.
The English founded second with Ascetism and Ceremonial Burial, having taken Tradition.
India third with Orders and Divine Inheritance, having taken Tradition.

Edit: I didn't manage to found at all. Maybe next time!

 

[SuperNoobCamper]

India seems to value orders a lot for a peaceful civilization that in my experience at least have never conquered a single city or declared an expansion war.