AI city placement and misc. suggestions

[civ4-advciv-oracle-bug]

edit2: nevermind i mean the AI was quite smart to go north (at least in save file 46.2, i played from a similar position, he went to plant north on the stone corn which i think was very nice), i was focused on ivory and such but he probably doesn't value them as much. I will change my strategy and plant closer to my borders on key ressources rather than trying to get everything and being too spread thin (on top of bankrupt plus border tensions), however as a general rule i think it may be cool to have AI more competitively plant if another player is threatening to plant with its settler close to the AI's settler and if most importantly that the land is good enough. But as this may be abused by humans players baiting settlers into settling by having a settler around, i'm not sure it's that good or important of a suggestion, but if you'd want to investigate it i keep the save file here i mean anyways, thanks,

Spoiler example 46 :

I have another example, it's a really missed AI opportunity.
It is example 46.

Here in save file example 46.1, Alexander AI's settler is in a nice spot to plant (there is wheat, sugar, and ivory and some hills, it's also a nice foothold i mean anyways to the nothern and west part of the map for further possible expansions).

Even if the spot is not perfect it is more than good enough, but most importantly a rival (of the AI) settler (i.e. my settler i mean anyways) is walking in this spot, thus i think it should be interpreted by the AI as as a planting threat.
It's really too bad that AI didn't decide to plant here, i don't know what kind of spot it envisionned by going north but letting me plant one turn later then he goes back very inefficiently after i plant. What was the point to spend a move to go north if it's just to go back south with no gain, while plantable good location was available during his turn. This is very inefficient and would be nice if you could make the AI be more enclined to plant if there is competition and the spot is good enough.

So in this save file just press end turn: you see the AI's decision to not plant despite my threatening presence of a settler.
Then plant where my settler moves to (move is already made)
Then end turn and watch AI go back very inefficiently, why go then go back with no gain of new city plant when he could plant a good enough spot during his turn.

Would be very nice if you could improve this behaviour i mean anyways for advciv if wished/possible.
Thanks,

edit: what's even weirder is that in example 46.2, doing the same actions but with the settler moving one tile east, then why the AI goes back and not north anymore? Why is the settler behaviour so inconsistent? Could be linked to the settler back and forth issues in other examples maybe (for example example 28)? Could you make the AI more efficient and decisive with its moves, in this case about settlers and planting decisions, if possible/wished would be nice i mean anwayys, thanks,

 

[civ4-advciv-oracle-bug]

I also have another suggestion related to this example 46 save file.
Here Rome AI is a bit too friendly despite me blocking his land expansion (my city 2 in particular blocks a few nice ressources he could have taken).

Could you make/add i mean anyways a relationship modifier "You're blocking our land expansion".
It's just an example suggestion, but it would work something like this :
"-3: You're blocking our land expansion" if less than 20% of plantable world has been already planted, and a player has borders with the AI, that would lead beyond the player's territory to significant unplanted area (that the AI could have planted if not for the player blocking the AI).
"-2: You're blocking our land expansion" if 20% to less than 40% of plantable world has been already planted and otherwise same conditions,
"-1: You're blocking our land expansion" if 40% to less than 60% of plantable world has been already planted and otherwise same conditions,
and finally if >60% i mean anyways of plantable land has been planted then this is not as relevant anymore (or maybe add a "-1: Your empire is much bigger than ours if for example as an example suggestion my territory is >200% of their territory? This modifier would be regardless of the land expansion modifier i mean btw anyways)

Additionally i mean anyways, since i took resources that could be his, what about adding a -1 "You have many resources, near our border, that we don't have" if say 4+ (in this example i have 8 ressources in my territory's range (including culture expand) within 4 tiles of the AI, 7 of which he doesnt have) resources are within 4 tiles away from the AI.

What do you think of these suggestions? Is it possible/wished for advciv? Would it take too much computation or maybe hassle to code, or not relevant for advciv you would think?
Thanks,

 

[civ4-advciv-oracle-bug]

so apparently the thread is here now i mean anyways

i have another example
It is similar to example 4
It is example 47

Here Augustus AI moves a settler group (with a few other units), which i think is not optimal because i am preparing to attack him and ideally he should be aware of this since i am his close neighbour.
I think a better strategy for him would be to bump his miltary and try to rush me into expanding his territory.
Spreading (and taking a few units with him will not be profitable until a long while), and unlike example 4 i am a direct threat to his cities (in fact i am preparing to attack him), so i think he should gamble and try to rush me rather than expand which will make invasion only easier (i assume this extra city will take a bit of a toll on his economy at least short term, plus he lost a few units to escort them there).
Would you consider making the AI more aggressive when its land is tight and there is a rival nearby, rather than expanding very far even if there is land?

(edit 2: about example 47, a similar behaviour happened again while he should be preparing for war he is spreading his units yet again, i think this is not optimal, i added save files 47.2.1 (when settler first appears) fast forward to save file 47.2.2 (after settler weirdly stationary for a few turns which is maybe not optimal but i don't know if it's the same as example 8's problem as he stayed only a few turns so maybe he was waiting for an escort i mean anyways) where anothe settler group (with a few units) makes the problem worse (or my invasion easier) as well as his empire weaker. If he is too thin he should prepare to fight me i think. Would you consider improving this too i mean anyways? Thanks,)

Thanks,

 

[f1rpo]

Yes, I had meant to write a brief explanatory post here as well. So the posts here up to incl. example 46 had originally been posted in the main AdvCiv thread. I felt that they eventually outgrew that place and asked a moderator to split them off. I chose the title of this new thread too.

And briefly on the subject of Monument culture: The culture rate of any building doubles 1000 years after its construction. Another poorly documented rule (of the original game). Difficult to give this more visibility – because the doubling generally doesn't matter much, so it's undesirable to draw attention to it. AdvCiv has already added a loading screen hint about it.

 

[civ4-advciv-oracle-bug]

Yes it's fine, maybe it can help me spam more (i mean develop more my examples xd), which i am happy to do, i just hope i do it efficiently and relevantly, in a way that is fun to me (i also like to be exhaustive ideally not repeting myself too much ideally but go in depth ideally i mean anyways) and perhaps you too or those who follow this thread.

Ok i see thanks, i thought it was a bug but i guess it's normal then. If you had to you copy paste a note on each culture building the civilopedia would get incredibly messy so maybe not a good way to handle it, plus if need to edit/change that later.

Also like i said before if you want to reply and give me explanations all good and welcome i would like it, just that if i had to choose i'd rather would like more to see code i mean (even if don't understand it in depth i can follow commit notes or something i mean anyways) or just reviewing the examples, it is just a mention and of course not for me to decide i mean anyways, i'll continue to submit examples that i find relevant hopefully as long as i want/like it i mean anyways.
Thanks,

I gathered the few bits of messages i mean anyways about example 48 here for clarity (deleted duplicates while doing so):

Spoiler example 48 (gathered bits of comments) :

About example 47, another thing i wanted to say is he made some nice decisions, he stole the pig from me by culture with one of his cities, and then he stole my silver by planting very boldly next to me, i was impressed i mean anyways.

However there is a major issue of him wasting all his available space went planting Antium grossly in the center, when he could have fit a few cities if only he planted on the edes of this tiny landmass. Doing so when landmass is tight may improve the Ai performance and would be nice to improve if wished/possible for advciv anyways. I detail below my reasonning and also mention the screenshot of example 48 i mean anyways.

I would like to raise these specific points in a new example (48) of when he planted antium or/and other cities, but as i don't want to spoil myself i'll only provide a screenshot for now with a mapping suggestion for cities which would be i think much more fficient/stronger for him as well as why below.

What i mean is i would like to criticize would be planting spots, he didn't exploit his land correctly and it would be nice if when land is tight to tell him to concentrate his cities more (same as example 40.2)

Also, i think there is another crab east so not only did he not use all his available land efficiently, but he also did not use sea ressources too, which i think is a big loss. I think a higher value should be given to grassland hills to not plant on them, for only only one food loss they give 3 hammers so they are generally very good value, rather plant on plain hills or other low food tiles if possible too. But my main point here is to tell the AI to compact its cities more if land is tight, i think it would make a big difference early, and the AI much stronger, up to size 7-8 approximately i mean anyways this city placement i did quick should provide a lot more benefits for him (in ancient era most cities are quite low pop so it's better to have more cities (say 3-4 cities of size 5-8 than 1 city size 11 no? Especially if land is tight i mean anyways). Could you tell the AI i mean make the AI plant more compact when land is tight? I think it would improve the AI significantly i mean anyways (and use sea ressources more perhaps). Thanks,

About example 48, i think it's quite important, as i said in the previous message i'll try to send corresponding save files when it's not a spoil anymore for me i mean (the save file when he plants Antium and/or his other cities one turn before, ideally when i met all civs and have world map if i don't forget hopefully i mean anyways).

edit 3: maybe a way to implement it is that if landmass is thin (say less than 6-7 tiles width), give a higher weight/focus/priority/intention (except locally much nicer to do otherwise for yiels or other things i mean anyways) on planting at the edge of the landmass rather than at the center. The reason is it's physically not possible to plant near the center city if <3 tiles distance so there would be much less cities than if they are at the angles of the landmass, planted on the coast. The other reason would be to better use the sea ressources and tiles, when land is tight coast tiles are not so bad with a lighthouse especially with an early moai as you said i mean anyways and/or if leader is financial

So would you consider changing these AI behaviours to make the AI more competitive, but also more realistic i mean anyways, would be nice if wished/possible for advciv i mean anyways.

Spoiler Additional example 48 bits of post(s) gathered in the spoiler below (and deleting the now duplicate below i mean anyways) for clarity, thanks: :

About example 48, now that i captured most of his cities, i can provide the save files for each city before it was planted (due to knowing when without spoiling myself of the world map or other things i mean anyways):

- antium founded in 2000 BC (save file 48.1)
- cumae in 1325 BC (save file 48.2)
- neapolis in 875 BC (save file 48.3)

I provided save files for each of these dates i mean anyways one turn before he planted them. If you want, you could investigate or try to make him plant cities on coastal locations rather to maximize the use of his limited and small land, and also to use the sea ressources (or tiles since space is so limited 2 gold per tile is always good to get i think anyways, maybe coastal commerce may be slightly better for him too?)

I think it would be most interesting for Antium city in particular to be planted on the coast, if Antium is not centered, then there is 3 or more tiles between cities which allow the other oens to be much more easily coastal too, but it would require playing (maybe skip turn very fast just to see his decisions), what i mean is i don't know how you would test but just what i mean is if he planted his cities on the coast it would be much better to better utilize and maximumly exploit/use/benefit from the available tiles if possible/wished for advciv i mean anyways?

i'll add an edit about that in example 48.

i also added an edit 2 in example 48: also added a screenshot 48.b that shows there was crab east too that was also a lost opportunity for the AI's potential city there i mean i think anyways as detailed at the end of the example 48's message. Thanks,

I added a 48.c alternative disposition in https://forums.civfanatics.com/threads/ai-city-placement-and-misc-suggestions.695343/post-16772115 (at the end of the post i mean anyways) which i think works even better, see the post for details.

edit:
about example 48, as said in https://forums.civfanatics.com/threads/ai-city-placement-and-misc-suggestions.695343/post-16772665, i have now advanced i mean anyways enough to provide save files for each date when each Augustus's cities were planted, to maybe investigate or/and make him plant on the coast to maximize space, as said in the linked post in detail. Thanks,

edit2:
also added a second screenshot that shows there was crab on the east side too as i remembered i mean anyways so it's even more of a loss for him i think, would be very nice if he makes cites more coastal when space is tight/not wide i mean anyways, especially when there are sea ressoruces he could have used i mean anyways, thanks,

edit3:
added a 48.c alternative disposition with 2 stronger cities and no bottom cities i mean anyways at least lower priority.
2 stronger cities would be nice if AI prioritized.
Looking at it, maybe as in examples 41.3 and 26 and 30 and similar food should be given a higher priority when settling, it would be nice if AI could exploit/use i mean anyways these crabs and hills instead of this center mess i mean that ruins any other possible worthwhile city spot in the peninsula i mean anyways.
Thanks,

edit4:
added another alternative version (2) with 2 stronger cities, with the upper city one tile west as compared to example 48.c, this allows the eastern city to be even stronger while the upper one can grab more grasslands (into farms) west, thanks,
note: in this example i planted cities more than needed, for example amorium will probably not be worth the investment of planting it or at least not until much later in the game, unless i'm mistaken i mean anyways, but at least i tried and learn maybe i mean anyways.

edit 5:
linked to example 24 (AI misses the corn nearby it could have settled near, see example 24 for details, thanks,

edit 6:
linked to example 95 where AI plants a city on a sheep grassland tile (nearby fish too), while a nearby location could preserve both sheep and fish unless i'm mistaken, see example 95 for details, thanks,

 

[Mr Smith]

Difficult to give this more visibility – because the doubling generally doesn't matter much, so it's undesirable to draw attention to it.

I think it's a great feature that should've been fine-tuned more, so that later buildings/wonders also have a chance to double. Wasn't there a thumbnail in RoM for it, adaptable to custom values?

 

[civ4-advciv-oracle-bug]

Spoiler edit4: about example 49, my mistake, i thought it was medic promotion but it is in fact guerilla, my main point would still stay but (much) less strongly, i think cities are not always planted on hills, but guerilla may be more flexible maybe too, i would suggest to enforce more city garrison for stronger city defense but i woul suggest this less strongly now i mean anyways, thanks, :

About unit promotions, i reposted it just below for clarity and removed old one, plus added an example save file as it happened again in this map i am currently playing yet again i mean anyways..
I added a save file example 49.

Spoiler Previous message about unit promotions (deleted the old one) :

I have another suggestion, it is about unit promotions.

For example in save file example 43, Incan AI has many longbowmen with medic promotion (i killed crossbowmen with medic promotion too from Boudica)??
While the idea is cool, another non defensive unit could do that
Also it's not efficient to defend less strongly then heal, better to have a strong unit and if it dies it dies, also a city provides base health regen anyway if i am not mistaken right?
So one medic of a not longbowmen type of unit why not, but to have so many medic promotions is just wasted city garrison.

The same could be said about city attackers and offensive units.

A doubly promoted city garrison is nothing like a double medic longbowman, applied to many units it can change a lot.

So would you consider hierarchising unit promotions.
- For archers and longbowmen, maximize city garrison before anything else. Whatever gains other promotions give for these units should not outweigh i mean anyways the loss of not having city garrison (About the archer rush, maybe make the AI opt out of it, because more often than not these are half assed for lack of a better word and not enough to kill someone so then the AI is in a weaker position after this rushed premature war i mean anyways, an axeman rush with city attack would be much much more threatening i think, since AIs get bronze working quite fast in the game i mean if i am not mistaken anyways, i would strongly suggest the AI to focus on this unit if it has to or wants to rush, these units could be upgraded later too so AI would be competitive with them more throughout the early and middle game at least i think i mean anyways) so i would use archers defensively instead for the AI if possible/wished i mean for advciv anyways). For crossbowmen and other more versatile archery units i would not be able to tell but please not medics as in this save file i mean anyways, maybe focus on most effective promotions so maximize strength promotion first?
- For melee units such as swordsmen/axemen, if person is a warmonger i would say to maximize city attack as it's so strong and would make a big difference (additionally to having them ideally be used to attack cities), or maybe one strength promotion and then city attack maximized? For more versatile units such as the spearman pikeman for example maybe maximize strength promotion first instead.
- For Chariots, horse archers, knights and more, maximize strength above all, i think it's better to have a strong knight that dies than a weaker one that has a small chance to retreat. The chariots and horse archers would be upgraded later into knights and cuirassiers and more, so i think they would be best with max strength promotion first before anything else. For mounted units like war elephants i did not use them yet so i don't have experience to say i mean anyways but since they can become cuirassiers i think having strength promotion first would overall be best or at least nicely serving.

On top of unit promotions, unless i am mistaken many times it seems AIs don't have barracks in their cities, so could you enforce them to build barracks before stockpiling units (since they are likely to get more units than the human player anyway especially early so better make sure they are strong).

Having control over AI's promotions instead of them being quite chaotic (i see a lot of medic and woodsman promotions which are more often than not quite useless i think (not talking about the free promotions i mean anyways but the chosen ones by the AI i mean anyways). I think it would make the AI much stronger and much more competitive by better handling unit promotions.
Would you consider this in advciv, what do you think of this (like i said i mean anyways i would be much happier than you review my example and implement code rather than reply to me and not review or/and not implement code, ideally if both are possible why not sure i mean anwyays, but if i had to choose i much much prefer you'd spend time rather on the actual code that i play against in games i mean anyways if wished/possible for advciv to implement i mean)?
Thanks,

In addition to what i said before in the message, in this example again (in the city of Ravenna of Augustus AI i mean anyways), while it is not nonsensical to go for medic promotion, it seems the AI likes very much the medic promotion on archers (and crossbowmen) from what i saw which may include or not other units i mean anyways. And again i think city garrison for defensive units should be much much stronger as a default
Also, it is a problem that many units are unpromoted at all, do the AI players have no barracks at all, or are they delaying their promotions based on context?

So regarding this, i would suggest again to enforce unit promotions hierachy, and also enforce barracks being built before most units, as the promotions always help in the longer run (why produce many units if they are all unpromoted, having a barracks before that should only help the AI i mean anyways).

edit i mean anyways: about the other topics unless i have something specific to say i will focus on the content of my messages rather i mean anyways, thanks,

edit 2: also i think woodsman promotions should be strongly discouraged for axemen and similar offensive units (units for which it may work quite well would be scouts maybe as i saw the AI doing or weaker units benefitting from terrain support, else i would focus on maximizing strength or their trait (city attack or garrison) for max effectiveness i mean anyways), whether it's for attacking or defending cities, there is almost never a forest or jungle where the battle happens (cities are never forested or jungled tiles i mean anyways), so generally default to going for strength should be much more helpful.
Unrelated but i personally think i mean anyways that woodsman promotion is way too weak, so i would suggest to buff it in some way, perhaps maybe i mean anyways for example giving it attack bonus as well as defensive bonus? But even then for a more general approach i would suggest to enforce i mean anwyays strength promotion in many cases as default for offensive units (unless melee city attackers like swordmsmen that may mix city attack and strength maybe), and enforce city garrison maxed out as default for defensive units, archers and longbowmen in particular i mean anyways as said in detail above. I think AI would be stronger this way, if wished/possible for advciv i mean anyways. Thanks.

edit3: you can look at example 41.2 for a crossbow triple medic promotion, very annoying, i mean it would be nice if rather/instead the AI chose actual pormotions that influence combat outcome, leave medic promotions for supporting/non-combat units, all the save files of the map i played between examples 24 and 44 could be browsed i mean anyways to see how often archers (among other units) have medic promotion or none. Examples between 17 and 23 should be all advciv 1.12 too unless i'm mistaken so you could also browse through these. Finally examples from 45 and above (current map i am playing at the time of me writing this i mean anyways) may also be browsed to see how often AI has this behaviour. It would be very very very very very very very nice i mean anyways if AI could have poromotions to its units much sooner (by building barracks asap in any game/city) and then choosing actually relevant promotions that are competitive from what i think i mean anyways (city garrison for archers/longbowmen, strength for mounted units, strength/city attack for melee attackers such as axeman/sworsdman, strength for spearmen and more versatile units) (and at best of the best if mounted units inclined civs, if they have a tendency to build horse archers/elephants/knights/cuirassiers, to have a stable before that would be even nicer, i think AI could afford it considering the big amount of units they produce, so having barracks/stable before they produce units (or most units) would be very veyr nice if wished/possible i mean for advciv anyways, thanks.

 

[f1rpo]

@Mr Smith: You're right, it's a BULL feature. Though even BULL disables it by default. When enabled, it shows e.g. "Double culture in 120 years" or "Built in 25 AD" in the tooltip of the building list. So long as the age of a building matters, there ought to be some way to see that age. So the option (to be kept disabled by default) would be nice to have in AdvCiv. Hm, it's just a single block of code in the DLL – but also the usual Python and XML shebang. I wish Firaxis had realized that idea in a more user-friendly way, perhaps a National Wonder, "Louvre" or something with "national heritage," that, once upon its completion, doubles the culture rates of all buildings that belong to any earlier era, i.e. going just by the tech requirements of the buildings and not their actual age – so long as they're not younger than the National Wonder. Or maybe something specifically for obsolete buildings.

 

[civ4-advciv-oracle-bug]

I also have example 50 very similar to example 43 (and other similar examples mentionned in example 43).
Lone legionary gets baited and attacks approaching units instead of defending city.

This makes the invasion a lot easier (his cities become much weaker now), so i think it really hurts the AI and should be improved ideally i mean anyways.
I think correct behaviour should be to focus and enforce ideally the AI to focus on defense if being invaded (unless it is much stronger but then it would be strange that someone invades him besides that, except maybe if his stack is attacking someone else far away and his inside territory is thinly guarded i mean anyways, in which case it would fall back to handling defense when attacking force is as strong or stronger, and in particular in all cases rgardles of strength to not get baited and solo suicide his units out of city garrison, neither one by one, nor in a whole stack either leaving city empty. The settling group with units was also sitting ducks as one of you said this word i mean anyways when i started my invasion, i think all this is (very) suboptimal and would be nice if AI handled this better even if a bit (ideally a lot) anyways), especially if military weaker or if a big stack is incoming, to not send their troops suiciding one by one but instead potentalizing the defense of the defensive stack.
It would be nice if AdvCiv AI dealt better at defending cities, it suicided so many times ruining any advantage it had, or making its bad situation worse, if wished/possible for advciv i mean.

(note: also unrelated but i gained randomly gems on hill grasslands which was nice in city 2 i mean anyways, i don't remember if it happened before in advciv 1.12 but it was nice to get a new resource especially one i didn't have and that helps a lot in happiness in all cities i mean anyways (with forges in particular))

 

[civ4-advciv-oracle-bug]

I also added example 51 which is also very similar to example 50 i mean anyways.

In example 51, a horse archer in range (unless i am mistaken) of retreating to a city as a defender (much needed) chooses instead to solo attack a stack. He dies while doing so, but even if he survived he would have been much more helpful in holding the city with the other defenders, potentalizing and i mean spreading the damage more evenly with all defenders giving a much higher chance of city survival. May not have changed the final outcome in this map in a few turns i mean anyways, but promoting such behaviour of the AI would help it a big lot.
Plus, i assume it was a newly produced unit, or some reinforcements, in all cases regardless why produce new units or bring reinforcements if it's to suicide them very inefficiently like instead of combining with the existing defense.
To fix this, make AI first retreat to cities during war invasion, and only attack if it can fallback to a city (or stay in a city if attacking from it), thus if it is wounded (and won it's good), defenders will defend it. Also attack only if odds are good enough, else it should be better to stay in city as a defender, first due to how rock paper scissor of units is handled favoruing defenders unless i'm mistaken, but also because city provides additional defensive bonuses (plus healing). So it would be much better when being invaded to have Ai stay in cities, or attack only if it can end its turn in a city, priroity being to stay in the city with the max number of invaders.
As said before, this assumes enemy invader is strong enough, else but why would the enemy invade i mean anyways, but assuming a weaker enemy (that can be defeated) invades, more offense should be possible if odds are good and only as a stack or in superior numbers to the invader i mean anyways, but still even then defense should provide better odds, just enemy may pillage so a middle ground would need to be found between always staying in cities when just one warrior pillages everything (avoid this outcome), and getting baited then slaughtered or ignored as city is invaded (avoid it too), idealy as i said only attacking in favourable odds not just when an enemy comes which is what it seems to be doing unless i'm mistaken i mean anyways.
As a general behaviour this may help the AI be a lot stronger (possibly also in AI vs AI warfare?)

Would be nice if advciv AI was capable of behaving that way i mean anyways if wished/possible, would you consider improving it? Thanks,

 

[civ4-advciv-oracle-bug]

I also added an example 52 similar to 50,51, but with workers.

For an unknown reason 3 of Augustus AI's workers leave his territory unescorted instead of staying inside cities (or at least somehow protected i mean anyways), so they are offered up to me for free near Rome, i could even capture (or attempt to capture i mean anyways) them with a warrior if i had (meaning any military unit would win)
I think correct behaviour would be to make sure workers in this position don't leave the city if they are not escorted, especially not if they move to a place where the enemy can attack them
Would you consider improving this for advciv too i mean anyways? Thanks,

edit: added example 57, very similar to this example, here i mean anyways: https://forums.civfanatics.com/threads/ai-city-placement-and-misc-suggestions.695343/post-16773577

 

[civ4-advciv-oracle-bug]

I also have an example 53 similar to previous ones.

About this behaviour, i don't think i would continue to post each of this specific example behaviour, but each situation is a bit different, maybe the multiple data points can help for tests or something i mean anyways.
Here it's a bit different because the unit leaves its duty (or what should be its duty to defend my incoming invasion) for a very unknown reason i mean anyways as i explain below.

Here an axeman in Ravenna leaves the city for a very unknown reason, instead of staying and expecting me to invade.
It's really too bad because he had +25% fortify bonus too, if he stayed my stack would have had a quite much harder time to capture the city i mean anyways.
Does this behaviour happen because of the woodsman promotion that somehow he wants to stay in the wood?
But defending the city is so much more urgent, what is there even to gain in the forest away from both me and his city?
So similarly as the previous examples, if war is declared on them and they expect to be invaded or could, have their troops stay inside their cities as defenders as much as possible (or stack patrolling perhaps if not too fancy anyways but not spread for no reason).
Now my chariot can get closer to his axeman and kill his axeman, making my invasion of Ravenna much easier than if it stayed in the city as defender.

Also, as i said in example 49 about unit promotions, it would be nice if he had a strength promotion instead of woodsman that would almost never help him significantly, unlike the strength promotion that should almost always if not always be helpful.

Spoiler my mistake about example 49, see edit4 of example 49 for details and what i would advocate (less) strongly on it i mean anyways, thanks, :

And as i said in example 49, if it possible it would be so nice if AI build barracks before any unit or almost any unit, and actually use promotions as described in example 49's detail post or in a way that maximizes efficiency (especially for archers, they are often unpromoted or weirdly have medic promotion (???), while the city garrison would make so much more ense and help so much more especially if a few/all archery units choose it), would make AI a lot stronger if possible/wished for advciv i mean anyways. Thanks,

Would you consider improving the AI units leaving cities and getting baited instead of defending which is much more valuable if possible/wished for advciv i mean anyways.
Thanks,

 

[civ4-advciv-oracle-bug]

i have another example 54 similar to previous ones but also a bit different.

Here a catapult gets baited like in the other examples.
Correct behaviour would be to have it stay inside the city to help as a defender.

The difference is that siege units could be handled even better by the AI, they could stay on "alert" as city defenders, and attack stacks that are next to the city as long as at the end of its turn catapult has enough movement to end in city.
While they have strength like other units to also help as a city defender alleviate attacking pressure of the incoming stack, additionally, because of their collateral damage, siege help not only in city as a defensive defender but also as an offensive defender: if it survives the attack its few strength points can still help alleviate/mitigate the pressure of the attacking stack by spreading their strength on many units rather than fewer units tanking all and more likely to die. To be more precise, having one more defender means strength of the attacking stack is spread more so each defenders tanks less damage and is more liekly to survive or/and kill the next attacker, plus on top of spreading the damage, one less attack attack means one less chance to kill a unit, influencing quite more than slightly the outcome of the offense, and if this behaviour is repeated by having no unit suicide but instead having the AI be smart and concentrate on defense (like i assume a human player would as much as possible if invaded, unless as explained below offensive defenders also attack while staying in cities which may help also or not, but due to rock paper scissors maybe defense is just better than attack even for offensive defenders i mean anyways), then the AI may be really competitive.

Similarly, it would be nice if other offensive defenders could, while they are city defenders due to invasion from an enemy, do attacks on incoming units/stacks as long as at the end of their movement they stay in a city, ideally in the city where the stack is closer to.

Would be very very nice if advciv had such a behaviour

Here in this example 54, what is worse and also why i added this example, is that the siege unit chooses to attack one unit, while it had the option to attack of 2 units. Why move with a catapult and not use its splash damage i mean unles im' mistaken anyways.

Would you consider improving this invasion handling behaviour as explained in previous examples, by having the invaded AI units stay inside cities, and some of the units that are offensive type (swordsman, axeman, siege etc) perform attacks from within the city they defend (to weaken the stack while the defensive units stay full strength i mean anyways if profitable, but edit: most likely due to rock paper scissors of defense it would probably be better if they stayed in the city without atatcking, and also due to fortify bonus per turn if i am not mistaken, on top of defensive bonuses of the city) or can move outside of their city defense as long as at the end of their movement/action they end in a city tile where an enemy is threatening to come. To prevent players (AI or human) from abusing this by pillaging everything while the AI is too scared to leave its cities, have some sort of middle ground where if the enemy is strong enough it can leave its cities and attack incoming units as long as the wounded units can (be strong enough + enough movement) all comeback the next turn to city defense without being ambushed and die there. Would you consider implementing this in advciv, i think it would make the AI much stronger if wished/possible for advciv i mean anyways.

To sum this up all, AI is much much more likely to survive the war and/or successfully defend its cities if all its units stay in cities when invaded (instead of getting baited out of cities and doing solo or stack suicides that can then be ambushed, making at the same time the city much weaker). The rest would be to handle pillage threats in a way that doesn't compromise i mean anwyays defense due to getting baited as a result, and also handling the possibility of chasing the enemy away of territory if strong enough, but one shold ideally assume enemy would not invade if not strong enough, hence falling back to handling city defense while invaded i think i mean anyways.
Quite realistic but would be very interesting i mean on top of ebing competitive of the AI i mean anyways.
Thanks,

 

[civ4-advciv-oracle-bug]

I have another example 55, unrelated to previous ones though i mean anyways.

Here a barbarian axeman attacks a city (Rome in this example i mean anyways) from a flood plains tile (city and this flood plains tile are not separated by river, this means he would not get a defensive bonus if counter attacked).
However, the turn before he moved to a tile from which he would be in range to perform a city attack the next turn i mean anyways, which was also a higher defense bonus tile: hills plains.
He could/should have i think ideally performed the attack from highest defense bonus tile possible.
By doing so, if he survives after attacking, he would have slightly higher odds to survive a counter. may not matter in this specific case or maybe would, but applied to many units and as a general behaviour would be nice to see AI do this for its own good too i mean anyways.

Plus, another advantage would be, if unit was mounted, since it was already on this tile, it could have retreated if surviving the attack too if i am not mistaken i mean anyways.

So all in all, i think it's a nice behaviour to enforce if possible/wished for advciv anyways.

An incoming human attacker would always (at least i think i mean anyways) try to maximize his chances/odds i mean anyways, he it would be by attacking (as much as possible i mean anyways) from the highest bonus defense tile allowing this.
Here in this example a barbarian performs this unoptimal behaviour what the barbarian does, but a AI civ player could have done this i mean anyways, would be nice if could be improved for advciv if wished/possible i mean anyways, would you consider improving this too i mean anyways?
Thanks,

 

[civ4-advciv-oracle-bug]

I have another example similar to example 53 and previous ones..
I try not to post them all, this one is slightly different too.
Anyways i mean anyways this is example 56.

Here enemy AI gets baited out of city garrison where it could have nicely stayed as defender from a city on hills with i assume also fortify bonus of 25% if i am not mistaken i mean anyways.
What is different and worse in this example is that AI attacks a stack much stronger than him (that is on top of it on a higher defense tile). It also fails to capture Rome (even if it won i was positionned to finish off its weakened units so really no profit except suiciding all his stack that was nicely defending antium i mean anyways), as i explain below.

In example 56.1, an AI stack is incoming, i don't know if it's to defend the iron mine from pillage, or just if (as usual i mean anyways) it gets baited out of city garrison, regardless of cause in all cases anyways i mean AI gets baited out of city. It ends its turn in higher defense tile.

Then in example 56.2.1, after i see the AI incoming stack on a higher defense bonus tile, i choose to retreat during my turn (on a higher defense tile)
At this point when i end my turn, correct AI behaviour should be for the AI to retreat too to city defense (to begin with don't leave city would bring no risk (there was no big benefit here except maybe protecting the iron but then don't chase the enemy too far away in low defense tiles, especially as seen later don't get baited to follow and attack the enemy that is retreating) as it would have acheived its assumed goal of defense while also having no benefit to further pursue me i mean anyways.

However after ending turn in 56.2.1, AI commits the massive mistake to chase me too far, and now blows/fails i mean anyways his offense, resulting in his stack getting slaughtered in 56.2.2.
Correct behaviour for the AI should have been to retreat.

It seems there is a massive problem with the AI invasion handling.
It almost always gets baited ruining its position or defense, which makes the counter attack from me into an invsion to be much much easier. AI should be content and focused on defense epsecially if weaker.

So what is different and worse in this example is that AI atatcks a stack much stronger than him.
Did its code break or something i mean anyways due to losing or being outnumbered, or did he fail to assess the best strategy which is to defend due to being invaded (on top of which weaker than the invader).

It would massively help if you could fix this behaviour of the AI ruinnig its chances every time almost i mean anyways.
An all-around "easy" or so it seems fix (but i don't know if wished/possible for advciv nor if it's easy i mean anyways) solution would be to always defend when invaded. Thus, cities get much higher chance to survive the invasion, and AI to be comeptitive in the game.

It would be nice for the AI to handle not only offense but also defense.
The getting baited massively hurts it every time there is war or so it seems almost always i mean anyways.
Would you consider improving this in advciv?
Thanks,

 

[civ4-advciv-oracle-bug]

I also have this example 57 very similar to example 52
Augustus AI moves 5 workers unescorted next to my military units.
Correct behaviour should be to move them escorted or have them stay in city, as regardless of what the purpose was they would be captured anyway this way (unless AI purposely suicided them to gain unit cost but seems unlikely as he could have just deleted them i mean anyways or better gifted them to someone else maybe if i am not mistaken i mean anyways).

I edited the example 52 to mention this one too at the end of the edit i mean anyways.
Would you consider improving this too in advciv if wished/possible? Thanks,

 

[f1rpo]

After capturing Ragnar's capital, i notice he has no granary (nor barracks) in his capital (no building at all in fact). (His other 2 cities that i destroyed were also empty if i remember correctly).

Apart from wonders of the world, most buildings have at best a chance of surviving conquest. The probability is building-specific and set in Civ4BuildingInfos.xml. Barracks never survive, Granary survives with 66% odds. Most culture buildings never survive either. I assume neither side is supposed to have them in a potential struggle for cultural control after conquest. Overall, the destruction upon conquest seems deliberately excessive with an intention of slowing down wars of conquest.

Available ressources are not improved at turn 200+, but instead have a weird fort on them.

Might be the same wrong assumption as in this recent post that Forts don't connect resources. On workable tiles, however, the AI should make sure to replace them with improvements that increase tile yields too. It's not a great use of worker turns in any case, but I think the priority is already pretty low. At least, this way, the AI has a few Forts somewhere. But I haven't looked at the savegame yet.

About relationship modifiers, [...] when an AI declares war on someone (whether it's another AI or a human player), the AI that declared war would get -1 modifier "You declared war on our friend" from the AI that got declared war on's friend(s). [...] unless i'm mistaken only the human player has such penalties [...].

The AI should get these penalties just the same as the human player. "Friends" meaning (somewhat counterintuitively) either Pleased or Friendly attitude. But I'll check in the savegames whether this is somehow not always working.

I like to address these (potential) misunderstandings as I notice them because this doesn't take much time.

Sorry btw to read that moderation delays have given you trouble. I don't think those existed when I first started posting. Hopefully, you'll soon have been registered long enough for that to stop. Though I really don't know what the criteria for that are.

 

[f1rpo]

Having skimmed through all your texts – and not having reviewed savegames or screenshots –, my impression is that the following subjects keep coming up:

[AI]
• general jumpiness, arbitrariness in AI behavior, esp. trade
• tech trades not sufficiently considered when choosing research
• imprudent interest in remote cities/ sites
• high-production tiles, high food-tiles undervalued when founding (specifically founding on metals)
• clusters of bad tiles not sufficiently avoided when founding, vanilla flat/hilly grassland underappreciated
• contested sites underprioritized, precarious sites not avoided
• tight surroundings settled improvidently
• problems with settler, worker escorts, choosing alt. sites when best city sites unsafe
• neglected safety of frontier cities (especially new cities) against Barbarians and rivals
• invasions lacking decisive force, predictably failed wars of aggression
• not turtling enough on the defense, pointless small counterattacks
• not targeting weakest city, failure to pillage
• choice of promotions not cutthroat enough
• slow to respond to rival getting ahead
• diplo modifiers not reflective enough of stuff that matters (geopolitics, closeness of borders)
• insistence on payment for peace
[UI]
• misleading or missing tooltips
[rules]
• revolt strength too high

A few general thoughts – not necessarily to repudiate those criticisms, which I can't fully appreciate without having considered the specific examples: I want the AI to err on the side of taking risks and long shots and interacting with the player as opposed to somehow just sitting there. Ultimately, it's the human player who should be and feel smart. Of course the AI can't follow this philosophy in a way that greatly reduces the overall challenge. Though I think one can also make the AI more inconvenient to play against without really making it more challenging. I wonder if a higher difficulty level would make minor misjudgments of the AI feel more like welcome opportunities rather than reasons to stop playing. Which isn't to say that bad AI decisions are in fact manifestations of cunning game design. And, if what the AI does clearly can't work, then it's only irritating. (You do seem to have an appreciation for "bold" AI moves too - if they can work.) The AI trying to use a variety of means (units, promotions, civics, wonders) rather than using almost exclusively the best also is desirable to me or at least not worth adjusting. The problem lies with the game balance then. It all comes down to specifics of course.

I'm getting the sense that, for many of these points, the number of available examples is now more than sufficient. The more egregious, the better, but maybe not the more, the merrier. Well, it's your call. For my part, I don't know that I'll look at all that.

 

[civ4-advciv-oracle-bug]

Yes, this is quite a great summary, thanks.

First to reply to your previous post, it is true that some of my questions come from misunderstandings or lack of knowledge. I enjoy the game as best as i can and how i like to, but sometimes there is stuff i don't know which i am happy or don't mind to learn about. Forts for example, i did not know that they connected ressources, however you'll see in the examples if you look that for example one gold ressource was not mined, it's quite the big difference especially in the early game.
In that sense also, i wanted to remember to remind you i mean anyways to also look at screenshots, they are not visible from the ingame client but may help a lot (i only provided them in specific cases where i found it relevant, mostly city placement or local specific situations that i wanted to show (edit: as you pointed my bad for not noticing i mean anyways) examples on screenshots, for the gold it would be example 41.2.

It is true also that i want AI to be the most efficient and play the strongest it can, this may mean some common strategies, however i also like variety and novelty too i mean anyways.

Spoiler my mistake about example 49, my general point would still stand with other examples maybe, as for what my mistake is see post of example 49 for details, thanks, :

However, in the case of the medic promotion, i saw way too many times the same archers defending cities with double or triple medic, i did not post it yet again but in the map i am currently playing AI double medic promoted the archer

. What i mean is that as you said, sometimes it's hard to enjoy the game when the AI does very suboptimal things. If it's about variety and such i do not mind, but if it's obvious misplays it would be nice if i could play more evenly against the AI.

You are correct to suggest that i may increase difficulty to gloss over or rather compensate for lack of a better word the AI fails with increased bonuses, but this comes at a cost, which is the game is more of a grind then. When i play against the AI i don't necessarily (in fact don't at all) want the game to become a micro management of every tile, i find that very not fun (edit4: i must clarify/take back what i said here or at least clarify it i mean anyways, it is not that i don't want to manage tiles, in fact i find that quite enjoyable it's fun i mean anyways, it's just the excessive need to hyper micromanage, including for example swapping tiles every few turns for micro gains is where it becomes tedious and not fun for me anymore, else i like optimizing my tiles with purpose i mean anyways so it was to clarify that i mean anyways, and take back what i said abruptly that may have been a bit unclear as to why i mean anyways, i still hate/find not fun/loathe hyper/ultra micromanagement of tiles in particular, but else i generally like optimizing my tiles i mean anyways or having a general plan or some occasional tile swapping but not excessive it's what i meant when i said then it's a grind and not fun anymore (plus i may not be best suited or skilled for this kind of things, maybe partly because i don't find it fun, i would prefer focus on strategy and game flow, anwyays, it is my small ramble and to also clarify too i mean anyways)) even though it is not very pleasant of me to say, however what i enjoy much is the strategy decisions i make and how the AI responds to them. I expect the AI to respond to me with an as strong strategy, be it in tech, city placement etc, to make the game more exciting and interesting. At least for its most obvious blunders and downfalls. Ideally, i would want to play against an AI that has no bonus and against which my strategic decisions really matter, but if it just a bit smarter on key points it is fine for me too i mean if it has moderate bonuses. You can imagine on highest difficulties how AI has way too many units and cities early, i would prefer a more even opponent that is smarter is what i mean, but to some extent i as you said think that it's fine to give it some bonuses, just not too much when instead it can be made smarter, which i would also enjoy more to play against by the way.

Another point where i agree with you is that sometimes things are just tedious. If AI is too good at war then conquest is impossible or an extreme drag. So harder is not necessarily better i mean anyways, at least it's what i think. But i think there are better ways to adress this issue than have the AI blundering its bonuses. I do not ask to make the AI perfect (even though i would much enjoy to see its play refined i mean anyways, it may give me insight on new strategies or way to play against it also which is fun i mean anyways), but some examples are really critical and would badly need improvement, at least i would much enjoy it. You can look, of course when you go through the save file, for example at example 37 where the blunder is very critical, to sum it up i mean anyways AI makes 14 of its units abandon the capital that has a few wonders and 100% defense, all to attack one single unit in an unroaded tile, and all the while i have a big cataphract stack threatening attack next turn with catapults with it. It is a big big blunder here as he could have easily held his capital or make it like a proper conquest if i have enough units, and the oversight would have been on me to not invade if i can't win, i would have been taught a lesson by the AI xd.
edit 3: and to continue on that train of thoughts i mean anyways, i would hope that maybe handicap bonuses may be revised if AI is smart enough to not need some of them (edit4: i don't know if it would be a good or adequate i mean anyways remark to make, but i think i read somewhere that kmod had less bonuses, if i'm not mistaken on its main description, what i want to say is just that i would feel good about a smarter AI being smarter in a way that it needs less bonuses to provided an adequate challenge i mean anyways), or at least that i would have to reduce the difficulty instead to compete with the AI (i would prefer the first but second option may work too if AI behaves differently/smarter).

Again i am very thankful of you going through all or most of my examples.

I was thinking of ways to hierarchise them (for example removing solved issues) but i thought leaving the linear order would be easier maybe to navigate, plus i would have to create an actual drive which was not my original intent, i would like it to be kept simple even though i would very much like to know if you'd find ways to make use of my suggestions i mean anyways, but if some are not helpful, at least i submitted them i mean.

For the future and as i said in latest examples, if the issue happens many times i will try to not post it, unless i think it's different enough to be of use to you i mean anyways. Also, if it's not, i'd rather post them than wonder, but generally i would try to post unseen situations that i want to ask even though i run the risk of them being possible to see or not if i think it is significant enough i would prefer to try to submit it, if it cannot be reviewed maybe it's not ideal or the best but maybe it's how it had to be but at least i did what i wanted i mean anyways.

Finally, i am thankful for your consideration throughout all this really i mean. Even though it is not for me to say i mean anyways, i think you are the best judge on how to reply or not or use or not my suggestions. I hope they can help and would happily see changes related to these, but ultimately this mod has a philosophy on top of being yours xd i mean anyways, so all i can do is really suggest and leave it to you i mean. The discussion is also interesting to me this topic and exchange so i feel motivated to go through what i wanted to convey. In the end, they are all but suggestions but i make them gladly or at least knowingly i should say i mean anyways.

Finally also i mean anyways, the issue with moderation was only with a few messages (something like 2-3 messages in all, and had not encountered it recently), i assume it was due (or one of the causes would be) the insertion of links that added extra delay for manual moderation, so i held from for example sharing the link to example 37. At worse i would delete the seemingly buggy or delayed message, but eventually it got through easily i would say, and if you did not read it immediately then the issue is fine. For now i can post directly even with editing or so it seems, it may not always be the case but it seems to be working well if was to provide feedback on what i can do on this topic/forum i mean anyways.

I expect it would take delay to go through all them, also to observe some behaviours you may use different save files of the same map i played, for example for the gold mine you can check why it was not mined at different times of the game maybe. Also, for some weird war movements for example, i have provided .7z archives that you'd need to extract too to show how they happen in the span of many turns, else you would not see them.
If you cannot or do not wish to go through some of them it is part of how it is and i can only submit what i want to draw to attention. I also hoped if you implemented a fix, having various but slightly different instances of the fixed behaviour may help for testing if it works and such but i don't know how you would want to do it and again it is just my suggestion.

With that being said, i hope you find the way easy to go through my examples, messages, save files, screenshots, archives if needed, etc, and would be looking forward to feedback or even better changes related to these, but whatever happens or doesn't happen is what it would have to be i mean anyways even though it may not mean much to say this or not i mean this is how it is, so thanks for going through that as it would take a lot and i would really appreciate, for what comes of it it would be what it comes down to maybe.
I don't know how to say it but thank you for taking the time and consideration to go through this all that i do i mean anyways.
In all cases thanks,

edit: about building loss chance i didn't know but i suspected so, however as i say in the examples what i would hope is for the AI to have actually promoted (even if just once) units early, as it is really unimmersive to see it produce dozens of units yet many are never promoted i mean, thanks,

edit2: i forgot to say, i tried to provide the save files to as close at possible to where/when the event i mention happens, most often if not always just the turn before the AI makes the decision. I hope it can help navigate the examples smoothly, but it would still a lot of time, so yet again i want to say that i do not mind not all topics given individual replies, and would rather (if i had to choose) that you focus your energy on reviewing them and/or implementing changes/fixes, but again it is not my place to say and i only want to provide my input or/and suggestion. With that being said, thanks for your consideration throughout all these messages and now all i can do is wait xd (while playing or doing other things i mean anyways), thanks,

 

[civ4-advciv-oracle-bug]

I have another example i would like to submit
I am aware i should not spam especially considering there are many to review already, but i think this one is significant enough to submit. It may or may not be reviewed as a result of the amount of the rest but it is easier for me to send it.
It is example 58.

Here America AI moves many (5+ if i am not mistaken) workers just to build one quarry (for stone).
The stone is in their territory (on the other side of the coast), but to connect it to their route network, they would need to build a few roads on my territory, here only one road, but the walking distance from their territory is quite far (10+ tiles approximately i mean anyways)
It is a problem that it is way too many workers (one quary and 2 roads) for that task, however the most problematic is that the workers then leave territory without connecting the improved ressource to their network (which would only require one road on a tile they are already walking on, and in fact walked twice on (once when coming to the stone tile, second time when leaving (sadly i mean anyways without roading before leaving this tile (hill)))).
As a result, unless i'm mistaken this means they have no stone despite spending all these workers turns on it.
It is especially a problem since they passed by the tile (hill) connecting the stone tile to the rest of the route network, all they needed was to do an extra road, which would have been instant (one turn) considering mant workers passed by it.

Save file 58.0 is the starting point before they start building the quarry (many workers gather near my coastal city to improve the stone in their territory).
Save file 58.1 is fast forward i mean anyways to when they are finished improving, and leave problematically without connecting the stone tile to the route network.
My suggestions on how to fix this would be:
- do not send excessive amount of workers for one task (1-2 max should be enough, route and quary on same tile would count as different tasks so 2-4 workers max for it for example)
- do not forget to connect improved ressources in their territory to route network to their territory, even if doing so requires roading in someone else's territory. If there is an issue with enemy pillage or pillage abuse as i have read somewhere about advciv if i am not mistaken anyways, maybe keep building the road with just one worker, or somehow affect relation modifier (if an improvement we built was pillaged regardless of where, i don't know if it is implementable)
- start by the closest tile improvements first. It may give the ressource a bit later, but avoids the inefficiency of walking on a hill twice for example when it could have been roaded first before walking to improve the quary tile. By doing so, then when walking back the tile would have already been worked (roaded here) so no risk "forgetting" it i mean anyways. In many cases, a small positive side effect may be that the process overall as a result is more efficient (no need to walk twice on an unroaded hill, it would be roaded the first time naturally) would be naturally faster. This would apply to ressources that require working tiles outside their territory, when in their territory they may do as see fit maybe. I don't know if it would always work but maybe it can help the AI this way.
- if distance is far (as in that case, maybe if walkable distance is say 5+ tiles outside of the AI's territory), do not send too many workers, keep them close (not too far from outside borders, nothing drastic i mean anyways like forbidding them to leave territory, not that i mean anyways) to the AI's territory to do more improvements per few turns instead of always walking. This may also prevent unwanted worker deaths if attacked.

Would you consider improving some of these points for advciv? The most important is of course i mean anyways to make sure the ressource is properly connected, which is not the case in this example unless i'm mistaken. Thanks,

edit: added example 81 where AI uses one more worker the next turn after starting an (uneeded) fort, to build this (uneeded) fort with now 2 workers, may be related to this with my suggestion of using less AI workers per improvements to increase efficiency, see example 81 for details.