AI+ v13.1

[terranu1]

War wise, AI civs usually don't really get that far against other civs but sometimes do make gains. City states are easily taken, but most AI vs AI wars end up in just a single city, or no cities whatsoever taken. War declarations are mostly appropriate, but they lack the ability to push through the unit walls. In what is kind of counter intuitive, some of the best conquerors I've seen were on lower difficulty levels, with the rare civ (mostly Rome) ends up taking over 1 or 2 other civs completely.
Against humans, they can be a little scary because peace orientated humans tend to have far less troops. But they're still easily outfought using simple tricks that allow you to hold them off long enough to break them. They still are only really properly scary at the very early game, or if you never build any troops. Sometimes a civ with mounted units can still do some harm, as can those that rush down the field cannon line. Their individual unit control is not the worst, but they do really badly at having coordinated goals (missing opportunities to take a city in a turn).
Defensive wise, they're still horrible. The main problem is still that they're unwilling to bring their units towards battle. Often because they're planning to attack some city while being attacked themselves. They build walls way too late because they all avoid masonry now for some reason. At least they sometimes get archers now.

.

Is there a way to force AI's to build corps and armies out of all their individual units when they got the civics for it ? Or to build only corps/armies once they have the civics unlocked ?
Or to make them upgrade their units ? (Had England and other blocking tons of sea tiles with 20+ individual archers whilst they got plenty of crossbowman or musketman corps mainland).
The only AI who seemed to make very good use of army civics was China who threw at me at least 15 AT crew armies..but for some reason still had archers and pikemans behind.
I tried upping alot the flavor for "build armies" or something in the files but it didnt seem to change much. If only they would do that, it would cleanup LOTS of space and make for a much easier planning when going to war.

 

[duozerk]

It appears that in the latest release, the settler, the worker and the slinger units have been renamed to "CORE END", "BEHAVIOR END", and "MILITARY END" (respectively in core.xml, behaviortree.xml and military.xml). Those may possibly be forgotten debug names ?

 

[V. Soma]

It appears that in the latest release, the settler, the worker and the slinger units have been renamed to "CORE END", "BEHAVIOR END", and "MILITARY END" (respectively in core.xml, behaviortree.xml and military.xml). Those may possibly be forgotten debug names ?

yeah, same odd names here

 

[V. Soma]

So, a quick test:

Ai+ v7, Smoother Difficulty1.2 , TCS Omnibus Mod 1.1.1 Complete Ruleset, 8 AoW,

Emperor, quick, 4-leaf map


Here are the stats after 88 turns:

Scythia declared on me and had big army (90% of army were horsemen),
yet did not attack my small city with horsemen, only with 1 warrior and only for a few attacks...

See stats, France went for OCC?

Cities:
Rome 5 (!), Scythia 2, France 1 (?), me 2

Buildings:
Rome 8 (!), France 6 (!), Scythia 4, me 4

Districts:
France 3 (!), Rome 2, Scythia 1, me 1

wonders:
France 1, me 2

Great People:
France 3 (!), Scythia 1, me 1

Religions:
Scythia 1, me 1

Well, Rome was powerhouse: science, culture…

 

[Siesta Guru]

Siesta Guru updated AI+ with a new update entry:

v7

- Removed debug names

Read the rest of this update entry...

 

[Siesta Guru]

Woops, forgot about the debug names, sorry! Fixed now. I knew this day would come..

 

[duozerk]

No worries, thanks a lot for all the time you're spending on this and for the new release !

 

[V. Soma]

If only combat AI improved a bit... just a tiny little bit... e.g. attacking cities...

 

[Puer]

Is this too-many-settlers issue still present?

 

[Siesta Guru]

Is this too-many-settlers issue still present?

If you're talking about the issue where one civ would sometimes get 4+ settlers, I haven't personally seen it in my last few tests, but since the underlying cause is still there it can probably happen.

If you're talking about too many settlers in general, thats still going to happen on marathon, but it should be mostly fine on the other gamespeeds.

So, a quick test:

Scythia declared on me and had big army (90% of army were horsemen),
See stats, France went for OCC?

Hmm, interesting.. Did France happen to struggle with barbarians? That's the only time I really see OCCs in my games.
Will have to look into the strange Scythia attack. Almost certainly one of those cases where she launched an operation that ran your way, declared war, then somehow couldnt actually find its way to the destination, or decided to give up halfway.

If only combat AI improved a bit... just a tiny little bit... e.g. attacking cities...

Yeah it badly needs something more. If only I could get my hands on the dll.. It's always going to be hard to do this stuff perfectly, but right now it doesn't even prioritize walking a unit on top of a 0 health city.

 

[Laurana Kanan]

I could also just give instructions to change a few of the most essential variables, but that would require people having to mess with their files. Any thoughts?

This would be much appreciated for those few(?) of us that play Epic & Marathon games.

 

[elWappo]

I could also just give instructions to change a few of the most essential variables, but that would require people having to mess with their files. Or maybe I should just keep it as is but warn that it's optimized for quick.
Any thoughts?

This would be much appreciated for those few(?) of us that play Epic & Marathon games.

I second this; I play Epic speed with custom SQL updates for much, much longer Tech & Civic times. I find this gives a nice representation of each era (much more time per era), while having relatively good units/buildings/districts build times.

 

[Siesta Guru]

This would be much appreciated for those few(?) of us that play Epic & Marathon games.

I second this;

Alright. I'll do some proper testing later and make an actual guide (or maybe make a separate file with instructions in it), but for now these changes should at least in theory make things more reasonable.

Find the following bit in core.xml

<Update>
<Set Value="9"></Set>
<Where ListType="DefaultCitySettlement" Item="SETTLEMENT_DECAY_TURNS"></Where>
</Update>

I would recommend changing the value here to:
Quick: 8
Epic: 11
Marathon: 14


Then change this (in core.xml)

<Row ListType="AllowExpansion" Item="PSEUDOYIELD_UNIT_SETTLER" Value="1.2"/>

Quick: 1.3
Epic: 1
Marathon: 0.8


Then this bit, also in core.xml

<Update>
<Set DefaultValue="12.3"></Set>
<Where PseudoYieldType="PSEUDOYIELD_IMPROVEMENT"></Where>
</Update>

Set it to:
Quick: 13
Epic: 11.7
Marathon: 11.2

 

[V. Soma]

Will have to look into the strange Scythia attack. Almost certainly one of those cases where she launched an operation that ran your way, declared war, then somehow couldnt actually find its way to the destination, or decided to give up halfway.

Scythia did come into my land with the horses, only they did not attack my city, only the 1 warrior that came along... and later 1 swordsman... the horses attacked my spears on the field and also my ally CS units... but there was a case when Scyth 4-6 horses could have attacked my small city...

 

[Lewi11]

Scythia did come into my land with the horses, only they did not attack my city, only the 1 warrior that came along... and later 1 swordsman... the horses attacked my spears on the field and also my ally CS units... but there was a case when Scyth 4-6 horses could have attacked my small city...

Minus the part where their horses didn't attack your city, did they have a decent battle? I.e did they actually attack your units and ally's units when possible or were there still a lot of units standing still\moving around while you fire arrows at them?

 

[Morgawr]

This is great stuff but the peace deals inbalance issue is very game breaking IMO. Coupled with AI being bad at tactical war, it makes short 10 turns wars too easy of a strategy to dominate the game in Deity.. I understand hands are tied until DLL is made public.. Nevertheless I will give this new version a try and see.

 

[Siesta Guru]

Scythia did come into my land with the horses, only they did not attack my city, only the 1 warrior that came along... and later 1 swordsman... the horses attacked my spears on the field and also my ally CS units... but there was a case when Scyth 4-6 horses could have attacked my small city...

These 1-unit / 2-unit groups people are sometimes seeing that behave a little different than the other others are almost certainly the units from my 'zerg' operation. They have their own (simplified) behaviortree and operation, and are meant to get more units into the warzone and do some light harassing without getting in the way of the larger troops. It does seem like these are mostly functioning appropriately, it's just that the larger operations that are also supposed to be there, end up dropping the ball massively. The bigger ones still mostly work on a vanilla behaviortree, which has proven to be very hard to expand upon to make it less indecisive.

This is great stuff but the peace deals inbalance issue is very game breaking IMO. Coupled with AI being bad at tactical war, it makes short 10 turns wars too easy of a strategy to dominate the game in Deity.. I understand hands are tied until DLL is made public.. Nevertheless I will give this new version a try and see.

Yeah I'd argue the peace deals are the single biggest failing of the AI at the moment. That it's bad at war is one thing, but that it just surrenders all of its assets whenever it does badly at war is unforgivable.
The only change I've been able to find that can positively affect this, is that I can set a diplomatic 'cost' and 'worth' to the peace making diplomatic action. Sadly testing so far suggests that it doesn't affect (or barely affects) how much AIs are willing to offer in their trades, and would at most affect their willingness to get to peace. But it's not the willingness where the problem is, it's them tossing cities at you like they're luxuries.
I've already added a conservative change to these values in v6, and it's still present in v7 because it at least doesn't seem to affect things negatively. Let me know if you happen to notice a change in this area, it's unfortunately very hard to test this, as I can't see it in my battle royale games (which may be why the Civ AI guy hasn't noticed).

 

[Morgawr]

OK thanks. Also, I hate to be a broken record about this but I started a new game w/ v7 and the lack of holy sites is still very apparent. Deity turn 60 and only *1* AI has made any progress towards a GP, and that's <10 pts. Other than that, only 1 campus and 1 encampment has been built in the whole world. The lack of holy sites is a stark contrast to vanilla - where there's a huge rush to get a religion before all run out in the first 70-80 turns. I don't have an opinion on how soon non-faith districts "should" be built, but I believe how AI prioritizes holy sites early game in vanilla makes sense and should not be changed..

 

[Siesta Guru]

OK thanks. Also, I hate to be a broken record about this but I started a new game w/ v7 and the lack of holy sites is still very apparent. Deity turn 60 and only *1* AI has made any progress towards a GP, and that's <10 pts. Other than that, only 1 campus and 1 encampment has been built in the whole world. The lack of holy sites is a stark contrast to vanilla - where there's a huge rush to get a religion before all run out in the first 70-80 turns. I don't have an opinion on how soon non-faith districts "should" be built, but I believe how AI prioritizes holy sites early game in vanilla makes sense and should not be changed..

It's not really so much that I initially changed much about how it prioritizes faith. It's much more that through the large web of changes I made in other areas, it can end up creating some new undesirable effects. What appears to happen now with faith is the same thing that is causing the city walls issue, some of the civs get the prereq tech too late.
The desire for faith and great prophet points is now actually really high before they get a religion, in fact, it is much higher than in vanilla. The pseudoyield ends up at a total of 6.4 on deity, with vanilla at 0.5, while the average pseudoyield valuation is also higher, it's nowhere near that much higher. On top of that, some civs have some extra desire on top of that value.

This issue is also a little frustrating in the sense that I'm getting feedback in terms of a relative lack of districts on basically every district type. AIs simply can't be competitive in all areas of the game simultaneously. Unlike in civ 5, these things don't scale well into higher difficulty levels because the AI is limited by the same district numbers on every difficulty. If all AIs would hyperfocus religion, then suddenly a culture victory may be super easy as they're all neglecting that. If they focus both, they'll fall behind on science quickly. If they try to keep science up too, they'll end up with horsehockey cities with 0 production. If they work on that too, they may have too few troops to fight off anything. If that is solved too, then were back to where we started as it's spreading so thin now that it's going to be mediocre on all issues.
A solution is to make civ-specific specializations, where each civ occupies it's own niche, which is kind of what I've been trying to get done, but that can also cause issues if the game is missing all of the civs that would normally fill that niche.

 

[Real SciPio]

...it seems that AI is so fixed on settling city spots(as hilghighted by the game itself) that it will keep sending settlers entire game until that exact spot is taken...no matter where that spot is!!
...seeing lack of farms...but too soon to confirm as a rule