Do you still experience AI military behavior bug in 4.21.1?

[Jeddite]

The easiest thing would be play out and see, but I am asking for myself and other people who might not want to burn several hours into a game that will have to be discarded/too easy/too hard dueto one AI snowballing conquests etc.

Situation is a little unclear as some people reported that behavior still happening in 4.20 thread but the GitHub ticket concerning it was closed way before 4.20 release if I remember. That's why I don't push it on GitHub again.

The question is not do you think AI is too powerful/too weak in army. It is not about the difficulty you play.

The question is do you experience clearly unintended behavior like AI leaving cities unguarded or retreating before you even if it has a larger military, or is not attacking you/defending properly when at war. If you exprience such behavior, it would be helpful to provide screenshots, I guess. I hope we can provide feedback, maybe the bug does not appear anymore. Let's say 4.16 is the reference version (the last one AI was working as intended according to many users).

 

[andersw]

I'm still on 4:20 and I see barbs doing the "lets move out with the full hp and place the injured barb in the camp" on occation.
AI moves out but at a time when it barely have any units anyway.

The only serious issue I see is that:
I see settlers unable to decide where to settle, my guess is their original settle spot were taken or barbs were near or at the spot so they are "afraid" and now the AI struggles to make a new descision, this leads into an AI getting halted at very few cities even as authority.
While I dont want vanilla garbage settle back, the AI needs to be ok with not perfect settle especially when at only 2-3 cities so they keep the tempo up.

 

[Alpakinator]

In 4.21 I'm experiencing very lackluster attacks across coast. AI knows what my army is and despite that makes an amphibious attacks across islands with a 4x weaker army that is destroyed in 2 turns.
However AI defends cities properly, so far doesn't leave garrison, replenishes it when possible, and swaps low HP garrisons for high HP when it can.

 

[Alpakinator]

AI also stupidly attacks cities with caravels instead of galleases. AI doesn't seem to know that naval melee ships are bad at attacking cities unless they have vanguard promo, or can move after attacking.

 

[andersw]

Howabout this in 4.21.1, zulu have 4 pathfinders around my expansion "El Jadia" at turn 79 epic speed.
I sometimes build one extra pathfinder, 3 extra says I've made military CS very happy.
And if I somehow would get 4 I would send two in different directions and sell the other two, definitely not go in a clump with them.

Spoiler :

 

[KungCheops]

I guess I made a mistake in trusting the existing "we need this amount of scouts" code when I did the unit production changes. Should maybe just hard cap it at 2 scouts at all times.

 

[andersw]

I guess I made a mistake in trusting the existing "we need this amount of scouts" code when I did the unit production changes. Should maybe just hard cap it at 2 scouts at all times.

There are some use for more than one scout to have vision over a large area and 4 scouts can probably clear a camp here and there but not the way I would use them and with the limited supply a human player have we cant afford that many.
With that said, a hard cap on 2 or 3 is probably good.

 

[Tekamthi]

I guess I made a mistake in trusting the existing "we need this amount of scouts" code when I did the unit production changes. Should maybe just hard cap it at 2 scouts at all times.

i dont think this is good idea, poly is finally playing right in my latest game, im unable to mass exploit ruins cuz AI has built appropriate number of recon for the map we're on. If you have to do cap, 4 is not unreasonable on larger map sizes in early-game -- maybe 2 is right on standard and smaller, but i dont like hard caps in this regard.

With that said, a hard cap on 2 or 3 is probably good.

2 or 3 is the right mix for most civs, most of the time, but i played recent game where i had 7/12 supply as recon in classical and used this to snowball a big lead. special case perhaps, but poly on huge map should have min 4 in early game imo. any civ with recon uu will not be well-served by these hard caps

 

[GeneralAmadeus]

So how about hard cap of 3 for most civs and a hard cap of 5-6 for some special cases like polynesia.

 

[azum4roll]

It should scale with map size. The actual dimensions, not the size ID.

 

[andersw]

i dont think this is good idea, poly is finally playing right in my latest game, im unable to mass exploit ruins cuz AI has built appropriate number of recon for the map we're on. If you have to do cap, 4 is not unreasonable on larger map sizes in early-game -- maybe 2 is right on standard and smaller, but i dont like hard caps in this regard.


2 or 3 is the right mix for most civs, most of the time, but i played recent game where i had 7/12 supply as recon in classical and used this to snowball a big lead. special case perhaps, but poly on huge map should have min 4 in early game imo. any civ with recon uu will not be well-served by these hard caps

Good call, I dont play with ruins.

 

[Tekamthi]

I dont play with ruins.

when this option is toggled, maybe AI could auto adjust to 50% of normal recon unit priority -- w/o ruins the case for ++++recon doesn't exist outside of uu war plans

 

[KungCheops]

I suppose the CvEconomicAI function for deciding how many recon units are needed can be updated to reflect these suggestions.

 

[Nudimmud]

I still have this bug, and I have it no matter what I do to the point I've stopped playing. The AI abandon their cities, even their capital, and a domination win is easy no matter the difficulty. It's a complete mystery.

 

[Agiwan]

I still have this bug, and I have it no matter what I do to the point I've stopped playing. The AI abandon their cities, even their capital, and a domination win is easy no matter the difficulty. It's a complete mystery.

Try this: https://github.com/LoneGazebo/Community-Patch-DLL/issues/11787

 

[Jeddite]

To anyone concerned: I am currently playing the latest version with the build from the post above and have been for a few days now.

I've been knocked out of the game turn 89 by Inca by losing a city as Poland in an otherwise flawlessly going game and being second in production and gold for quite a while. Deity, progress, he had two cities against my five, composite bows were just always too few to both deal enough damage to force him to withdraw two units a turn, there were always three of them engaged on me side, rest healing or on their way, instead of five or six. I mismanaged timing.

This is how it should be, the build is great. AIs and barbarians are performing strong and menacing.

 

[Recursive]

Due to this bug, I will create a new release fixing it.

 

[JNZS13]

Try this: https://github.com/LoneGazebo/Community-Patch-DLL/issues/11787

To anyone concerned: I am currently playing the latest version with the build from the post above and have been for a few days now.

Wow... I had stopped playing a while ago because of this bug and a version that fixed it was ready days ago? Many thanks @ilteroi, gor always imptove tat ai and thanks @Agiwan for show us the link.

 

[Nudimmud]

Try this: https://github.com/LoneGazebo/Community-Patch-DLL/issues/11787

The game won't start at all when I use this... what could possibly be the reason for that? :|

 

[Jeddite]

The game won't start at all when I use this... what could possibly be the reason for that? :|

You have to just replace the same file inside Documents/ balablabah/ Civ 5/ Mods/ (1) Community Patch, I assume you did it.

Then probably mod-mods that confict with changes in this version? But it should be crash at mod-mods or game loading, not game start. It'll be included ine the version that will probably be uploaded in few minutes/hours. you might just delete everything and apply new version.