AI city placement and misc. suggestions

Little by little you keep reinventing my modmod😀

Just providing a lot of culture that will also double after 1000 years is enough to justify any early wonder, especially for Industrious.
Like i said for the most part i was mostly looking for little tweaks to enhance the performance of the AI, or/and make the map more interesting strategically, but without making it too complicated, adding too much variation for each tile may not increase the game experience for me, but it may be true that some of my suggestions align with yours i mean, just that i would favour simplicity and strategical depth/local position that global changes to make it "fair". I detail this a bit more in my below message (if i dont forget hopefully i mean..). If anything i would prefer it to be less fair but with smarter AI, making it more strategically challenging without being a grind, ideally i mean.

As for the wonders, i really think for most wonders culture should be a side effect and not its main bonus, some +100% culture wonders make sense, but most should (i think) focus on the effect on the wonder and not the bonus culture gained (and great person too i mean). The effect of each wonder would be nice to improve i mean.

So all in all i would advise for simplicity and depth, but not complexity that is vague, if that makes sense i mean, and making AI smarter/stronger without buffing its bonuses (ideally reducing them proportionately to the smartness gained) would be nice.
 
About example 4, i understand that it is indeed not a bad spot that provides many resources, i would still favour the western one with corn and cow i mean, but i understand better that AI may favour it. However about the barbarians and their pressure, i would say planting too far south would not supress completely the pressure (unless AI uses one of its units to "watch" unseen tiles near the city as a human would i mean but it's probably overkill or too much hassle to code if ever possible, would be nice though if AI did this but it's maybe a bit too hard/impractical to implement i mean i dont know, but considering the number of cities it loses to barbarians in almost in each game i play, centering planting in a radial path rather than longitudinal one, when possible may help overall land security i mean. But it's true that not planting south may result in even more barbarians, so it's a tricky position that no player would immediately contest. Then the approach may to tell the AI to not settle too deep uncontested (by any player) land unless it has a few units to garrison it. Often these cities are destroyed due to one unit guarding the new city. Maybe telling AI to concentrate a bigger/most of its units closer to its borders (especially ones next to no other civilization where barbarians could spawn) would improve the AI if not already implemented by advciv? If not already existing, would advciv consider implementing a ay to tell the AI to guarrison more its outside cities, and when planting near no civilization use 2 units instead of one? (which hopefully shouldnt slow expansion too much? But it may be cheaper to lose the settler/city once in a whiel than wait everytime until 2 guarrison are ready (especially in the early game where expansion is crucial i mean i think) so maybe it's best as is and unfortunate i dont know, but anyways what would advciv think of this?)

Another consideration about example 4 is probably mitlitary security vs other players too i mean: if AI plants this spot of corn and cow, the human player or another player is more likely to contest it, so maybe it's best to not go for it overall i mean

About ressource visibility, it's cool i didnt know, it means that AI cant just blind find but really needs to think about which spots are overall best, hopefully this knowing good spots strength would translate thorughout the game when choosing where to plant and possibly when too but i dont know about that last point. The interesting thing about planting spots, in relation to how you said before i mean, that other saves files may provide other examples, it turns out i mean i have another example from today, example 13.1 and 13.2 (same city spot just at different time locations to show the surroundings etc i mean). Here AI very outrageously plants on the copper grassland tile, that would have been so nice for city growth otherwise (it's like 4 free hammers and no food cost (plus copper as a reosurce but planting there also gives it so not the main point i mean)). I have provided screenshot and save files for this example 13 here (download link at the begining of this message: https://forums.civfanatics.com/threads/advanced-civ.614217/page-80#post-16754101 ). This time however unlike example 2 city was planted in 1440 AD by Zara Yaqob who has Bronze working since a very long time, also unlike example 2 this is not a hill iron but here in example 13 a grassland copper.. I feel it's really a waste. I understand the strategic value of the fresh water and coast, also very far there is silver, but considering the very long distance from capital city (unless there is another city in between that i dont see (didnt want to spoil myself also i dont remember how to switch to AI view would need to reread your post but it looks a very cool feature but also no spoil for me now anyways), what i mean is i think the other spots i pointed nearby also achieve freshwater or/and coast. Also, like in example 4, if worse comes to worse, it may be fine to not get either fresh water or/and coast bonuses, but i think the value of this copper grassland tile is really higher. Hence, i am asking if this intended advciv behaviour, and wouldnt it benefit the AI in that case to plant in one of the spots i highlighted? (again if no other city or obstacle is there that i did not see). If it is intended, wouldnt it make AI in advciv stronger to not plant in strategic resources (especially if one grassland for production resources, or most tiles for food ressources)? So as in example 2, would advciv want to incorportate these changes? Unlike example 2, Zara Yaqob doesnt seem cramped this time, and can go for the silver later even though it doesnt look like surroudning tiles are nice enough for that, but i would favour not planting on key resources when they could give such nice yield, 4 hammers have been, i think, wasted in this example. What do you think of this, and would advciv consider changing the AI's judgment on planting on a resource depending on surrondings and the tile it's on or/and other parameters.

About the rest of your message, i did not read all yet, but i also agree that things like boosting tundra is in fact quite unintuitive, and may not appeal to most. My idea was to make the game a bit fairer by valuing each tile, but there's no harm if some are useless. In fact, it would make the game faster if the civs are cramped sooner, it's why i choose high see level by the way to not have the game drag for ages, one can always choose a bigger size if one so wishes. I also understand and learn from your explanation about advciv cramping civs more in map generation unlike vanilla bts, for the same reason as in this paragraph, i am in favour of that, even though the game mod is not made just for me xd, but i personally like when it's faster and not too large to settle everywhere so that strategy and stuff happen sooner i mean (else slow gamespeed relatively or something but it's not my point i mean). I just wonder about one thing, even if these tiles like desert, tundra, mountains are quite useless, isnt/is there another to make them useful besides settling on them or near them. Mountains have a very cool thing of not being able to walk on them, so they are like walls that add strategical depth especially regarding military. But is it realistic. Up to some eras, didnt civilizations sprout in mountains, and some cities (despite atypical) exist in these? For the sake of immersion, but also (my main point xd) not ruining nice planting spots unplantable due to it being a mountain, how about the suggestion to make mountains walkable (if not a hassle to code/implement i mean and if advciv would want too i mean) past a certain era? This seems quite practical and realistic too i think. For example, how about allow walking and working mountain tiles (building roads etc) when Enginerring is discovered? With the movement gain +1 i think it makes sense. In example 2, a mountain tile blocks access to all the nothern peninsula. Had the AI access to it, with its big army Boudica could have probably stormed the barbarian city there, and become much stronger, Boudica had a big army with nothing to do of it, plus was cramped, so it's a good example i think/feel of this suggestion. Another is as i said planting spots that are often too ideal but theres a (or many) mountain(s i mean anyways). So what would advciv think of this? If mountains are walkable, then settling on them is not too far fetched right? How about either this eing possible at Engineering, or maybe another tech? If another tech would be the way to go to allow settling on a mountain, what about a tech that follows Engineering? At this stage of the game settling would be just a small bonus anyway, and unless local circumstances, just a way to make the game easier/more comfortable i mean not giving critical advantage. Looking at the tech tree actually, maybe allowing to walk on mountain and work the tiles at construction makes sense, and settling on mountains at enginerring makes sense i think, what would you think of this suggestion? Again it would be immersive as you said, and personally realistic, it's not like it's unwalkable i mean, and sometimes it may help, movement speed may be reduced if really want to make it more realistic, but i am not sure it's required especially if the tile is roaded i mean. So what would you think of this suggestion in advciv, personally i would be looking forward to it. As for settling, in order to not haave full mountain cities everywhere, a defensive bonus of +25% same as hills should be fine and fit in civ4? What do you think of this suggestion i mean?

As for the rest of your message i will read and reply to it a bit later hopefully not too late but cannot guarantee that it would be too soon, in all cases thanks,
 
i have not finished reading and replying to your previous message, but i encountered today a very aberrant AI behaviour which i think would be nice to enhance if possible i mean.

It's quite related to example 8 (the Hammurabi save files scattered due to me forgetting to group them in a folder and rename them with "example 8" tag i mean anyways..), AI aberrant settler behaviours, but this time the opposite.
In example 8, AI waits a long turn with settler in city (maybe for barbarians to go away or to have a stack to kill them, however there is a nice plantable spot south east i mean ignored for a very long time by the AI as i detailed before in the example 8 first message explanation i mean.

This time in example 14.1 (https://filebin.net/rd9yfus93oui4aes), when you end turn AI moves 2 military units to escort a settler and military units during wartime. This happened a few turns after i destroyed a big city of Zara Yaqob, but the AI trying to go to settle again during war time only makes conquering the city so much easier, i can kill those 2 units out of the city (without defense bonuses), also the settler is dead. He should have known i have a quite big stack coming (maybe probe with one unit such as the chariot first before moving the settler?). In all cases correct behaviour should have been to keep his units in the city, building a settler during wartime is also very questionnable, but if it was already there, still the settler should be protected. As for the military units, they would help a ton guarding the city, which i may have failed to conquer otherwise. (Example 14.2 just shows the current situation after ending turn in example 14.1)
(edit: i added example 14.0 the turn where i destroy their city which may be the trigger for the settler event of example 14.1 and 14.2 to see what the AI does i mean)

So based on these 2 examples, would AI moving settler behaviour be enhanced in advciv based on these 2 examples? My suggestion would be settle if it's safe do not wait, if a desired spot is taken dont wait and move to second best choice especially if it's closer to other players, even more so if it's a nice spot, else, if it's not safe, do not go settle, especially during wartime where to garrison cities should be a priority (or go on the offensive if higher military power and that no units are threatening the AI).

So what would you think of these in advciv?

As for the rest of your message i will read and reply some time later or other time hopefully, thanks,

edit 2 : actually it seems that this roaming settler during wartime behaviour happens regardless of city being destroyed. After replanting the destroyed city (in hopes of having no revolt), i noticed revolt percentages are the same, even if i move a tile away the percentages are still high. If that's the case anyway that i get a revolt, then i'd rather keep the high population and religions spread inside this city already, and slightly better reputation with Zara Yaqob (by not destroying the city i mean anyways...) if i make peace later.
So i reloaded before city being destroyed and kept the city this time, however AI still escorts a settler during wartime, instead of concentrating his troops, allowing me to easily kill this settler. I think it's really inefficient and AI should keep its settler in cities protected, unless it wants to move to a safe place. Plus, being undernumbered, he should concentrate his troops not spread them and become weaker per city. Finally, i mean if he is coming in my direction and my armies, better not come and die losing the unit and settler for no benefit, however if going away from me (for example to survive and settle away from me) then it would make more sense though, not the case in this game. So would advciv want to make the AI not suicide its settlers and escorts during wartime or dangerous times/zones. I added the example 15 save file in same download folder. Thanks,

edit: added example 67 similar to these examples 14 and 15 if it can serve as an extra data point to test or observe else it may be ignored maybe i mean anyways
 
Last edited:
Hello,
I did not finish reading and replying to your comment @f1rpo, however i encountered a new AI situation that may need improving or would be nice to improve if possible i mean.

it is example 17 in this same download link: https://filebin.net/rd9yfus93oui4aes

In example 17, AI declares war despite the war being suicidal, so after declaring war AI retreats just after invading.
It's quite silly behaviour.. It's smart i mean of the AI to retreat instead of suicide, however if that is the case that the war can't be won, then do not declare war in the first place.

If they based themselves on my military power rating being lower and then suddenly discovered i had many archers with city defense promotion, then maybe have one of their units scout my cities (or scout the city that will be invaded) just before declaring war would help a lot. Would advciv consider improving this?

Btw, i played this with the 1.12 advciv, it's fun i mean that i mean i played this version before noticing your comment about a newer version being available xd. So far it's cool and AI behaviour is quite consistent, in fact i had to reload due to them wining the war and change my strategy. It's also cool to know that you'll want or/and will look at saves files i submitted. I hope they can improve the AI and make the experience more interesting hopefully i mean. I have other save files of older or newer dates throughout the games i played if needed i mean. Thanks,

edit: in this example 17, i find very bold and really nice that AI planted angkor city here, there's a lot of desert but it's a quite place to have a foot to tunnel to the rest of the land, also it's on river, founded on plain for least tile loss, it's far enough from my closest city to avoid culture influence or something, yet it is not too far from their closest city, so i dont know if this is the result of some changes in 1.12 about AI settling logic but i find this spot really cool. Yield may be a low but should be fine, but strategic position is very good i think. Too bad AI didnt go for the corn though but it's a bit far so i dont know if there was a better planting spot here i mean for the AI. What do you think of this i mean?

edit 2 : i have another example (example 16, still based on advciv 1.11) similar to this, i would call example 17 "ascertain war is winnable before declaring/starting war", and example 16 "ascertain city attack is soon before declaring war". I attached save files in the example 16 archive, from 1200 AD to 1360 AD. Since the behaviour happens in this range maybe the intermediate files would help. The big problem in this example 16 is that AI (Boudica) declares war, invades with 16/17 units (i dont know why one unit went missing (from 17 units to 16 units i mean anyways, coincidence that it's the same numbers as the 2 examples in this comment but anyways...) between 1200AD and the next turn but anyways i mean anyways...), and just before attacking city, AI army stays stationnary for some while (a few turns if i remember correctly) which is quite weird and not efficient (would cost gold maintenance too i mean anyways), then AI army retreats making me hope/think maybe my stack was big enough to deterr him from attacking me/maybe he switched target to someone else, but in fact he comes back in 1360 AD with a huge stack of 36 units if i remember correctly and at this point the game is lost, at least a few cities. Would it not have been better if AI waited until it felt it had enough units in a way that the number satisfies him to start city attack directly after invasion, instead of starting war, hanging around aimlessly, retreating, then comijng back veyr inefficiently (on top of being misleading me xd but anyways i mean anyways). Would advciv consider improving this too i mean? I think it's quite related to example 17 that ahppened in advciv 1.12 i mean. In all cases thanks.
Btw it's interesting to note i think i mean in this example 16 that my vassal AI runs away instead of defending his city, making me suffer all the losses, maybe if he stayed city could have survived? Would you think making the AI vassal fight harder if city is defendable or has a chance to be defended be a good improvement too? I cant be sure but he had a few longbowmen if i remmber correctly and some nice units, not sure it would have been enough but it sure would have helped a lot. Would you think making AI vassal fight harder when city is defendable instead of running way be a nice improvement too? (Plus it's his city to begin with!! i mean anyways...) In all cases thanks,

edit 3 : about this example 16 too would it be possible to tone down the revolt mechanic a bit? I ask this because to begin with i mean i halted my expansion partly due to having to suppress revolts. However due to not knowing how they work or how best to reduce them, i ended up after a few revolts making my stacs station in the conquered cities of my now i mean vassal. But if the revolt mechanic was toned down a bit, i could have expanded sooner and attack Boudica because Boudica attacks me i mean. This is espeially true because Boudica was fighting several wars which would have been a nice time for backstabbing him i mean. I also timed poorly my attack and maybe did not suppress revolt efficiently maybe i mean, but still it seems quite unrealistic for a city to revolt for 500+ up to 1000 years and still have a high revolt chance (i dont remember the exact numbers but 50+ turns if i am not mistaken) which if i knew i would have just let the city riot endlessly and push instead. It seemed that the percentage of culture incease very slowly (is it what triggers revolts when garrison is low?), which was espceially slow because it seemed culture was halted during revolt, making it even slower to culture convert the captured city (is it hat halts revolts?). So i would suggest toning down revolt mechanic a bit, but maybe it is because i did not understand correctly how it works and how best/most efficiently to adress it, in a way hopefully that doesnt slow too much my expansion while still ebing fun to me i mean and realistic i mean maybe anyways. I would lean towards toning that down, but i may be mistaken or have a bit harsh judgment opinion maybe or maybe not, so what do you think of this i mean?

For the rest of your message i didnt finish reading and replying to i hope i can do so at a later time but that may not be too soon i mean but that hopefully i remember/find time/will to do i mean if i want i mean hopefully. Thanks,

edit 4 : fixed missing files of example 17 i forgot to upload it seems i mean, it seems i forgot to upload example 15 before so i fixed that too now i mean
 
Last edited:
Today i have also encountered a very weird behaviour

It is example 18 (attached in same download link as in the message above).

I thought that the hut drop was deterministic based on the seed and that reloading the save would always produce the same outcome if i understand/know it correctly i mean.
However, if i am not mistaken i mean anyways, in this save file i mean if you press explore on the scout (before ending turn), you get from the hut points toward the tech Mining, however if you reload the save and instead walk manually doing the exact same move you get another scout. Is this normal/intended

Shouldn't the same action at the same turn produce the same result regardless if it is an automated explore move or a manual move?
edit: renamed example 18 save file for a clearer name but it is the same savefile

Also, another unrelated i mean question i would like to ask, is it possible to add the option in custom game to start with a worker? In higher difficulties AIs have one or many workers anyways if i am not mistaken right? So would the option "Start with a worker" be added in advciv in custom game i mean. It would help tackle higher difficulties i mean, or/and perhaps offer a slightly different gameplay and type of game that i may want to try too. I am aware i noticed i mean that it seems to be possible in the world builder, but it spoils the map or/and ressources and is not as convenient i mean, i may do it if really needed though but if possible i would prefer a specific option. A parallel option would be "All civs start with an extra worker", so that even in lower difficulties it would be fair maybe. Would advciv consider adding this custom option? Thanks,
 
Last edited:
Hello,
If i understand correctly, i would have to add a tag like this one:
<FreeUnitClasses>
<FreeUnitClass>
<UnitClassType>UNITCLASS_WORKER</UnitClassType>
<iFreeUnits>1</iFreeUnits>
</FreeUnitClass>
</FreeUnitClasses>
For each civ right? (i did not find the erainfo file by the way i mean).
If so it would be quite tedious, but the main problem would be to have to toggle between 2 XML files based on whether i want this setting or not, so although doable i would much prefer to have a togglable option ingame in custom game mode options.

It would be very cool if you could compile it, however in the future as advciv changes wouldn't it have to be recompiled each time? If so, i may want the longer approach instead of asking to implement this option directly in advciv so that it sticks in all beta or stable future builds. I am playing advciv 1.12 so i dont know how often you'd have to recompile it (if you want latest features i mean), so i'd like to ask first if the option could be implemented in the mod rather. Unless somehow the option can stick in the game independently from advciv developments, but i don't know enough about how this works to know i mean.

Thanks,

edit: another problem with the XML edit of this tag for each civ would be that if i want only me to have an extra worker and not all other civs, then i would have to each time i want a civ edit the XML accordingly, so i would be toggling between 4 XML files that would do :
- default (no bonus workers for anyone)
- 1 extra worker for each civ
- 1 extra worker only for the human player (i would have to update it each time i want a new civ, plus it would not work if i want a random civ too i mean anyways...)
- combination of option 2 and 3 : 1 extra worker for each civ + 1 extra worker only for the human player (so in total 2 workers for the human player and 1 for each other player)
So it seems easier to implement these 2 options if possible i mean, thanks,
 
Last edited:
Maybe iStartingWorkerUnits in GameInfo\Civ4HandicapInfo.xml is easier to work with. I think that'll apply to all players (AI and human) that play on a given difficulty level – whereas the iAIStartingWorkerUnits value of the human handicap should (additionally) apply to all AI players (regardless of the difficulty level that they play on). Unless a scenario (WBSave) specifies a particular difficulty level for an AI player, all AI players have Noble difficulty as their "player handicap".

Regarding determinism, automated exploration essentially uses the same code as the AI for exploration, which will often (depending on the situation that the unit is in) employ random numbers from the game's main RNG. And that affects all subsequent random outcomes (such as goody huts). Automated exploration is just about the only way that a human player can deliberately use up a random number in almost any game state. I've found this handy for testing purposes and also, on rare occasions, when I badly wanted to get a different combat outcome after reloading. (To be clear: auto-explore is already randomized in BtS; not my idea.) Come to think of it, one could use the Python console instead:
CyGame().getSorenRandNum(1,None)
The key mapping for that console seems erroneous when not using QWERTY as the system layout. On my German keyboard, I have to press Shift+ö to open the console, which I only figured out by trying random combinations. Parentheses, mercifully, are simply at their QWERTY positions, i.e. one to the right of QWERTZ. And this requires the "chipotle" cheat code (maybe for the better). Still, perhaps this is good enough, and most players don't know the auto-explore trick anyway. (Or do they? Would it be missed?)

That being said, while I can easily find the one or two places where auto-explore usually uses the RNG - and replace that with some unpredictable hash value, guaranteeing that automated units don't ever use the RNG would be a taller order. If I'm not going to establish such a guarantee, then perhaps better not to bother and to leave things at "automation may involve randomness."

That aside, dialing down the impact of randomness for human auto-explorers might be a slight improvement. So that the unit will mostly just moves in the direction that makes the most sense. Automated human moves don't need to be as unpredictable as AI moves. Unpredictability is arguably not desirable at all for automation – although a bias like going north when all things are equal could be undesirable.
 
I see thanks, so if i want to i'll see how handicap XML settings can help to try the "human player also starts with a worker at higher difficulties", since they have bonuses, it would make it less of a grind, but i assume if i need the bonus worker perhaps higher difficulties are still too hard for me maybe, the emperor bonuses were a bit too much in the middle game so for now i'm going back to monarch.

About the randomization of auto exploration ok, i didnt know it was an expected behaviour. However, i still see value in reducing this randomness. Just like the bonus hammer on a plain hill, it may be a bonus perk that maybe doesnt harm but is not especially meaningful either. Put it another way, there is no disadvantage to remove or at least reduce it, so i would lean towards fixing it even if partially, as long as it's not too much of a hassle and is wished for advciv i mean.
About the console i will not meddle too much xd at least for now, but it is cool to know that it exists i mean.

Similar to this randomization point, sometimes the randomness in fights is really crazy and perhaps a bit too random.
It would be very very nice, if there was an option in advciv, if i create a custom game, to have less random fights.
I understand that randomness to some extent has some purpose and is not nonsensical, after all in history fights were not strictly deterministic, but sometimes it's really ********, for lack of a better word i mean, warrior wining against an axeman, or many unexpected wins on a row.
As this randomness is part of the game and i think enjoyable to some extent i would not want/ask it to be all removed, but having a "less combat randomness" may be something really interesting, if it's possible to implement i mean, also since it would be an option each would be free to use it or not. I don't know if it's possible technically or if you would wish it for advciv, but since it's something i would very much like to try, then i am asking here.

So what would you think of these suggestions? Thanks,

I have a message that is in review for quite many hours, it's for me to handle this but just to say i may delete it and upload again as it may be a bug, but i'll see it myself i mean a bit later i mean.
 
previous message too long to moderate that may have been a bug or not but anyways here it is:

i have another example related to example 16, it is example 19 (examples 19, 17, 18 (and most likely onwards examples too) use advciv 1.12+).

AI declares war then if you just skip turn twice AI retreats without attacking city.
If such is the case that i am strong enough that a city attack would be suicide for him, then he should instead not declare war, instead of declaring war then retreating after declaration of war.

I attached the example save file in same download folder as the previous examples: https://filebin.net/rd9yfus93oui4aes

So it would be nice if AI doesnt declare war unless it's determined to actually attack me and thinks it can win.

Another problem is i think maybe strength evaluation function?
We had open borders and i think he is close enough to see inside my city (capital) before declaring war, therefore shouldnt he already know i am stronger than him defensively and not attack.
So would it be possible to have AI better estimate chance of attack success, the relative military power is quite misleading i think, in fact i am slightly stronger than him i think in this position with more units and city barricade, also promoted archers, and axemen that counter his swordsmen.

Also, maybe attack another city instead. After retreating AI is willing to talk and wants me to give up city of Susa?? Why would i bother to defend xd if it's just to give him a city later. Since i spendt energy and hammers to be as strong as him, he can't expect me to give him a whole city i could easily defend myself in this savegame i mean. So no, i mean since i am as strong as him, his peace treaty should be fairer, perhaps a bit of gold or something at most, but i dont think he has currency, neither do i, so maybe just peace then? Having a hanging war for no purpose doesnt seem optimal nor helpful (he pays gold per turn for military support for the invasion too if i am not mistaken too right i mean?).

Finally, i mean, could you make in advciv the AI target the weakest city instead? Since he wants Susa city so bad, why not just go for it directly instead of attacking my strongly guarded capital, he has the advantage of the initiative so cant he better plan his invasion? City of susa that he seems to want is not as guarded, and in other situations (other maps i mean) such an algorithm/logic of targetting weakest city first (even if just to make peace soon after) would be more effective i think. If this logic already exists, perhaps enhance it/tweak it if possible i mean? It would be nice if possible in advciv i mean to have the AI more competitive in war tempo i mean.

So, i mean similarly, after this finally i mean, would it be possible to make AI more likely to make peace after it has made enough gains (for example gaining 1 city or 2 cities quick), and war is likely to stall or be quite much longer i mean? It would be good for the AI too i think, as a preemptive peace treaty may allow for another war sooner.

Also, i mean after this finally too i mean, i notice that the AI that declared war on me has no reputation penalty of "You have declared war on our friend!". However, if i declared war on him, i would have had such a penalty. Is it fair?? I mean, i think the AI should get the war declaration penalty, unless it would mean to say that none of these AIs consider me their friend, despite having open border for quite some time i mean? Thanks,

edit: i also added an example 20, AI thinks it cannot win the war (just after starting it), but while retreating he could pillage for free (no penalty for him) my sheep tile, would advciv want to enforce pillaging when it is nearby and "free" (no penalty for the AI) to do so. It would make me weaker, and my units are 2 tiles away, so even if he stays here one more turn i could not catch up to him (and at worse he would lose only one unit, so he could pillage with the archer if he prefers, even though in this example it would not change anything as AI is too far from any of my units). Would advciv consider implementing this change i mean? In another try much earlier in turn number he pillaged everything due to me being too weak which i think was good, but now he retreats without pillaging which i think is a lost opportunity for him. Thanks,

edit 2: added example 69.1 similar but AI does not invade at all (see respective post for details), thanks,
 
Last edited:
I have encountered another situation that i would like to ask about.

It is example 21, attached in same download folder as the rest of other recent examples i mean: https://filebin.net/rd9yfus93oui4aes

In this save file, if you capture the enemy's city with the axeman, enemy city has 7 workers, yet of all those i only captured one?? I don't mean to be too strict i mean, if i am not mistaken advciv has reduced the worker capture chance (was it 100% in vanilla bts?). Still, i would expect something more like 50% chance, having like 3 workers out of 7 would seem fair, but 1 in 7 i mean isn't it a bit too low? Was i just unlucky in this example, or is it around expected probabilities?

It is not unusual than in other advciv savefiles i played outcome is more often than not "destroyed" about when i attack worker. I understand that worker capturing is maybe a bit too op, but isn't the chance a bit low here? Would you consider upping the percentages (or reducing the randomness of the luck if it is random?), or is it fine as is for advciv as you intend it? Thanks,

edit: added example 86 with a bit too high worker capture chance (5 in 8 in this example) and a suggestion to maybe make the worker chance calculation "formula" or whatever it is called i mean anyways i don't know the exact word i mean but it to tend more often towards averages, see example 86's post for details i mean anyways, thanks,
 
Last edited:
I also have another question, example 22 in same download folder.

AI is asking gold for peace while clearly losing, i have tons more gold, military much stronger, and captured a few of his cities in the few turns since war started.

Why is he asking for gold when he should be giving gold to me since he is losing? Is it working as intended.

I also remember havng a similar issue in a few advciv games (in fact unrelated but in civ3 too where i am wining but AI makes more demands for peace somehow). This behaviour is not realistic, would you consider fixing it i mean in advciv by making AI make realistic peace demands? (in this case it would be the AI either asking for a free peace treaty without gold on my end), or paying me gold to spare him. Is he accounting the distance between me and his other city (which i have no idea where it is, and perhaps he thinks war penalizes me more?). Since he is losing, i would expect me to either propose a free peace treaty of give me something to spare him i mean as i said. What do you think of this, would you consider tweaking it or is it as intended or not modifyable easily? Thanks,

edit: i also added example 64 similar to it but worse (as in the AI asking even more gold relative to stage of the game even though he is (quite) badly losing)
 
Last edited:
I also have another question, it is about example 23.1, but a similar question also happens in example 23.0 so i provided these as 2 separate time chunks if you want to investigate one or both i mean i think or the 23.0 for context maybe i mean anyways...

It seems that AI got weaker between both wars it declared on me, without any other player attacking him. Shouldn't the AI focus more on rebuilding military units, (even if slaving 1 or 2?), especially after a defeat, to possibly comeback or at least survive. It's not like my stacks were big in each war just about enough but he could have easily survived with just a few more units say 4-5 units maybe produced in his distant cities and brought closer to my border (since he has only Russia that didn't look from what i saw about to attack him i mean so he cold focus on me more).
Possibly also spread his troops so that they are more in the edge cities and less in the center cities?

The wars and peace timings are :
-war in 700 BC (he declared so end turn to trigger it),
- then peace in 450 BC (i made the peace treaty request, but i didnt save it so load 425 BC save file or declare epace manually from last 450 BC save file)
- then war in 325 AD (i declared),
- then peace (i) at 425 AD,
- then 700 AD (i).

Would you want to make advciv focus more on the military production after a war, especially after a defeat? I mean from a logical point of view if i attacked him once or he attacked me once it is more likely than average that a war between these 2 players happen, either for retaliatin (take back the lost cities/revenge) or finishing someone off (take a few more cities and maybe defeat/vassalize the weakened player). Also, if that is the case perhaps more offensive units than defensive if AI is military stronger too i mean anyways...). He had time and enough cities to produce enough units between 425 AD and 700 AD, yet somehow he seems even weaker. Did he face tons of barbarians or was his economy collapsng, but if so why did he keep planting new cities, did he not expect me to retaliate later and went more expansion route? Would you consider making the AI more wary of a war (and prepare offensively/defensively for it) if a war happened before (perhaps also if there is quite the power differential with a neighbour)?

It seems quite similar to example 1 where AI stagnated for a long while in tech never researching Scientific method for any of the AIs for many many turns which made me stop this save file.
In this example 23, AI seemed weaker after first war, why is the elading military power not making any gains, why doesnt he maintain his military advantage and redeclare war later, he had swordsmen so all he needed was to produce them. Would the economic strain be then too big to win? But he already had more units so just a bit more and he could have attacked anyone. Would you also consider implement this "If we have more military units, we are more likely to produce a few extra and declare war on someone sooner to capitalize on that (offensive units such as the swordsman i think i mean). He could have easily crushed me or another AI with just a few more units after first war, yet at 2nd war he seemed weaker. Do you have information about that from my savegames, and feedback about it, as well as would you also want to implement the other logic i mentionned of "if a war happened, another wa is more likely to happen sooner (retaliation or finishing off), so we prepare for that by producing more units sooner (offensive or mix of offense/feense based on relative military power (in fact just 2-3 more archers/longbowmen maybe be a threat enough to stop any invader i think)", but instead he just stagnated and even regressed hile he was the military power and could easily (i think) keep up and attack soooner. The last would be "make our military station more at the edge and less in central/distant cities, if there are several players focus more on the most dangerous one (or most likely to attack based on relations maybe?)" (maybe he had more units far and i just didnt see them? if so they were useless and he could have mobilized them sooner if they were ready near the border, keeping only minimum for garrison happiness or barabarian defense if need to check barbarian possibility too i mean), also the "we want more offensive units if we are stronger, more focus on defensive unit (especially in edge cities) if we are weaker)". If such things already exits, would you tweak them so that this situation doesnt happen and AI is ready to defend or attack when the opportunity/time is good?
This behaviour of AI being (unexpectedly) weaker after first war also happened in some of my other advciv games i mean (in 1.11 though) if i remember correctly.

Thanks,
 
Last edited:
I found an example of optimizing a city plant spot, it is example 24.
Save file is in same download folder, plus there is a screenshot too in same download folder.

AI secures some nice ressources, but misses the corn. By planting just one tile up i mean north i mean (spot 1 in the screenshot), then AI would get same benefits (gold, silver, grassland hills, many floodplain tiles) but with the corn which is much higher value.
The advantage of getting the corn is much bigger than the disadvantage of planting on one flood plain tile and losing it instead of planting on plains tile (which would be otherwise nice, but there is corn which is much better here i think i mean) so it seems like a big missed opportunity for the AI.

So far in the small sample of planting spots i saw in 1.12 i mean i dont have much things to say about the planting spots by that i mean they are really good i think, but here it in example 24 it could have been much better.

Would you consider implementing/improving this in advciv i mean, to prioritize nearby food more (especially when there is a lot of production tiles nearby like grassland hills, gold, silver etc) it would be really nice if possible/wished in advciv i mean.

Thanks

edit: looking at the screenshot i mean the only reason i can think of is he wanted to get the silver bonus sooner and not wait for culture expand, however i think planning long term in this city spot choice i mean would be much better, but if such logic already exists perhaps tweak it so that advciv would be more aware of the food nearby which would help the city a lot. Since just a monument or any religion is enough to eventually get the borders, i feel/think i mean that it would really be much better to plant one tile north i mean.

Spoiler my mistake actuallly AI handles that well :

edit 2: maybe another tweak would be to make them search mysticism sooner if they have a nice city spot but the ressources are in the outer ring i mean, so they could build a monument, especially if they have a chop nearby in city spot i mean, would be nice if advciv would take that into account too i mean but i dont know if it's feasable not too hardly or quite easily, but if possible and you'd wish to implement it i mean it wold be nice, ragnar would be slightly less aggressive early and get mysticism soon so that he also gets a monument and corn in this nice city spot i mean if possible/wished for advciv i mean. In all cases thanks i mean.

edit 3: perhaps the ideal time to research mysticism would/may be i mean after bronze working since they tend to research it quite early it seems i mean, or around that time maybe i mean, so they'd chop a monument as soon as they get it in relevant cities (in other cities use the chop for more important strategies maybe like workers or settlers or granaries for example maybe i mean but is just my opinion i mean). As for me in this game it's unrelated too i mean but my strategy was not good and i'd need to change it i mean but it's me to handle that i mean was just to be exhaustive i mean anyways.

edit 4: with this much extra food slaving is even more efficient i think, i didnt consider it, so maybe mysticism is a nice tech to have soon or quite soon after bronze working if AI has nice cities that need 2nd ring and no religion immediately (since religions are more costly in tech price + also in relations if others have other religions, while slaving a monument is cheap and fast, especially if there are some good food tiles in 1t ring i mean to reach city population 2 sooner i mean anyways), so would you make the AI value mysticism more after bronze working and perhaps slaving a monument sooner if no religion is adopted soon. I think that you'd most likely already hve considered this but i ask just in case, i think it could improve the AI to get just mysticism and focus on monuments and not necessarily the full religions too early, combined with better city plants that take into account the 2nd ring more perhaps (but most likely this already exists and AI "missed" the corn in this position? But maybe not, so i ask anyways i mean to know if you have the answer and would like to explain to me i mean i would like to know anyways), so would you consider improving this i mean?
(unrelated note: by rushing bronze working i found a better strategy to get the settler and spot 1 just before they plant it, but is for me to handle anyways it was just to be exhaustive, so i'll continue from now but just would like to know if you'd consider make the AI focus more on the corn in this position as in spot 1 i mean, and that it may hopefully maybe i mean apply to other similar positions in this map or other maps i mean, making the AI stronger maybe), if it's possible/wished for advciv i mean. Thanks,

edit 5: actually nevermind these last few edits i mean, it seems AI handles that very well and get the borders expanded soon enough, in fact sometimes or often sooner than me, so i'd just focus on making the AI aware of the corn or/and bearby food in similar planting spots if they can also keep the other valuable resource (else maybe trade-off prioritizing food or happiness based on a formula i mean or soemthing, just a suggestion i mean anyways).

edit 6: linked to example 48 where AI plants on a central thin landmass (city of antium i mean anyways), missing (and eliminating due to minimum distance between cities i mean anyways) the possibility of settling on the coasts rather to get both crab ressources, see example 48's post for details, thanks,
edit 7: 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,
 
Last edited:
I have another example similar to example 19, it is example 25 (in same download folder as the other files).

AI declares war but then retreats before attacking (most likely because he thinks i am strong enough to defend and he would lose more attacking).

I think this is quite inefficient, shouldnt the AI better evaluate if it has a chance to win before declaring war?
For example, in this position i think (i am not sure i mean) he can see inside my city to see i already have one archer, based on that, should he assume other archers are on the way and possibly i mean cancel or delay his war declaration (possibly gain a few extra units?)? But doing that may make the AI too reluctant to attack in this and other more general circumstances i mean maybe.
Should he for example declare open borders and then have a scouting unit while still stockpiling on his weakest aimed target (while doing so adjusting if i build too much units to change his target), basically delay a bit the early war declarations that are often preemptively cancelled by not going through with the attack but instead withdrawing. If he built a bit more units and gathered more info about his targets, maybe his attack would succeed more? (assuming he doesnt delay too much else the other players would get strong enough to defend). Because these war declarations end up being a flop and perhaps he should be more decisive to have more gains, the lack of tempo makes me aware that i need to bump my military, maybe other AI players (i dont know) also take actions based on this (failed) war declaration (for example producing more science or buildings since they are not the target, or plotting a backstabbing since he will be weaker and less defended eventually, or/and spread too thin i mean).
So would you consider making the AI ponder more its war declarations, so that they are more effective, and that the suprise effect is preserved for maximum impact maybe (i mean as much as is doable i mean)? Possibly this would involve a bigger gamble in having a few extra units in order for the invade to win, but then i am not sure if the price of producing all these units and to spread them thin in many cities would be worth in the long run, i would say yes due to taking care of a rival early but there would be some period to stabilize this expansion and build back the economy, but should be worth in the long run no?
What do you think of this, and would you consider the AI reevaluate its war declarations so that he doesnt do this weird war declare then retreat before attacking, which doesnt advantage him and is quite not immersive i mean. What do you think of this suggestion i mean?

Thanks,

edit: after playing this save file for quite many turns, he goes back and forth with his army, losing i assume gold per turn for unit supply right? Shouldnt he be more decisive on whether he intends to pillage or attack or leave territory to not lose gold, instead of hanging around for many turns in same tile far from city attack position as if he was brain dead/freeze xd i mean (or the weird back and worth that is not effective nor efficient i mean)? I have not attached save files for all that range due to not having played it yet and also not sure i would do for this example i mean, but it is similar to example 16 (that was in advciv 1.11), where the AI hangs around after war declaration and does weird back and forth very unefficiently and purposelessly or so it seems, either attack cities or pillage or not lose unit supply (and keep surprise effect of the invade maybe until all the full force is ready in example 16 for example i mean). In example 23 a similar behaviour of back and forth without attack of his stack before war declaration (i did not provide save files but can if need) and during war if i remember correctly also happened until i killed a few of his troops to chase him away. So would you consider tweaking this logic to be more efficient in war declaration and handling? I can provide more save files (that i have i mean) if need i mean. Thanks.

edit 2: added example 69.1 similar but AI does not invade at all (see respective post for details), thanks,
 
Last edited:
I have another example related to prioritizing food when planting city spots, it is similar to example 24, it is example 26.
I have attached the save file and a screenshot in same download folder.

Here Ragnar AI wants to expand east ward i mean but he could have planted just 2 tiles away to get access to deer instead of this very low food location. I don't have all the info on his west city i mean and dont want to spoil myself, but since the sheep is already pastured i assume it is in use by the city using it. If that is indeed the case i mean, why not plant in the spot 1 i indicated i mean to get access to deer?
Just like the corn in example 24, it seems AI is not aware or doesnt value nearby food ressources, but food is very very important for city growth pace, slaving etc, so would you consider make AdvCiv AI city plants focus more on nearby food? Thanks,

edit: also there is 3 mountain tiles (so unusable/affectable to population i think if i am not mistaken i mean), making it an even poorer location on top of the abundant tundra and no metal or production. If it is for the fur he could get it too while taking dear too, this city spot feels very cramped on top of being poor, so would you consider focusing more on food for city plants in advciv i mean? Thanks,
edit 2: also assuming the sheep is already in use by the west city i mean wouldnt it hamper the growth of the west city? I dont think AI juggles between tiles so better maximize food from land ressources instead of split them right? Thanks,
edit 3: are mountain tiles counted as very low yield of food/hammer/commerce or counted as neutral? Because unless i am mistaken i mean unless a special event happens for a mountain tile it would otherwise not give any yield at all right, so shouldn't mountain tiles also be penalized for settling (1 to 2 max is fine though but here on top of low food several tiles are mountain i mean). But mountain tiles are sometimes necessary in an otherwise good spot so i am not sure they should strictly be penalized else the good spot goes away for no good reason/alternative benefit. So maybe increase further the value of good tile yields i mean (if i understand correctly the word yield that i apply it to food and commerce tile value not just hammer i mean) rather than penalize (too much?) mountain or/and bad tiles? Maybe a practical approach of making the AI in both examples 24 and 27 settle in a way that values the corn and deer in these examples would be more effective than many theories or assumptions i make i mean maybe. What do you think of these ideas i mean? Thanks,
 
Last edited:
I have another example similar to examples 24 and 19, it is example 27. The situation in this example also happened in other situations maps if i remember correctly i mean so it seems like a reincuring behaviour (for example example 16 in advciv 1.11), which would be really nice to fix.
I attached the save files in example 27.1 (the actual save file) and example 27.00 (not the main file but just to show ragnar is ok with open borders at that date i mean (when i discover it i mean) as explained later in this post).

AI declares war then withdraws one or a few turns later before city attack.
As explained in example 24 and others maybe too i dont remember exactly my explanation i mean, but it seems he estimates me to be strong enough so his attack would fail or be too costly.
But if so why attack in the first place, why not gather a bigger stack, target a weaker city (but i could just move my troops then or target his undefended ones so maybe not a good idea i mean anyways).
In example 16 (advciv 1.11) AI does back and forth and eventually comes with his full power (while my vassal abandons me instead of dying with me in his city!! which could have maybe survived with his help, as i explained in detail in example 16)
In examples 19, 24, and 27, he hangs around aimlessly (losing gold per turn of unit supply i mean), and either retreats after losing a few units, or pillages around after many turns of hanging around if i am strong enough to defend my city but too weak to attack his stack (this behaviour is good/fine i think as he gets some value for his invasion and gold per turn loss of unit supply i mean anyways).

So would you consider as in example 24 and such other examples i mean i gave/explained here, to make the AI more decisive about its war declarations (which would be for example to either get a bigger stack and full invade city, or stop half way the military rush to focus on economy instead, city expansion of new cities i mean anyways, or perhaps switching to a new target, or maybe other ideas i mean anyways)?

One interesting thing he did is he proposed open borders (example 27.00 attached, a bit later soon before war he directly propsoes open borders which i think is interesting but i could not pinpoint easily exactly which date i mean he does it so i provided the first time since it's possible instead i mean anyways), but regardless of if i say yes or no, he declares war anyways, if it was done to scout me it would be nice, at first i dropped my guard in a bit different save file assuming he finally wanted peace due to seeing i am strong enough and stomped him in first war, if he proposed open borders i mean that i thought he would want finally peace i mean, so it was nice catching me offguard with it. Should open borders agreements come with a 10 turn peace treaty? Else it can be broken stupidly xd like he did unless i am mistaken of how open borders work i mean.

What do you think of this? I'm aware you may not reply immediately or too soon, but as long as you do it would be fine i mean, not too late ideally, but it's how it is so i would be content i mean i guess i mean to have a reply eventually especially if it is detailed and thorough i mean, plus going through the other examples which would take time too i mean, but i hope they would help me, possibly others too, i mean anyways. Thanks,

edit: if you play from example 27.1, you'll see that AI eventually comes back with a bigger stack, same as in example 16 (advciv 1.11), i provided example 27.2 of when this decision to reinvade with the bigger stack happens, so why not directly invade with a bigger stack? Or do other things as i said before i mean in this post i mean anyways. Thanks,
edit 2: in this example 27.2, you can see AI withdraw at the end of the turn, then reinvade next turn which is really silly i think unless i am missing something, couldnt he merge his stack directly? It feels/seems like he is conflicted/confused i mean by 2 contradicting logics/orders and does and undoes his moves every turn which is really inefficient on top of being not immersive i mean, so would you consider and have ideas on how to improve this war initiation handling and similar related things/ideas i mean as i said in the message i mean. Thanks,

edit 2: added example 69.1 similar to example 27.1 but AI does not invade at all (see respective post for details), thanks,
 
Last edited:
message taking weirdly too long to moderate so i am reposting it after delete here i mean:
(edit: probably due to the link, after removing the link then the message was posted directly)

Thanks, i thought it would be too spammy at some point but it seems @f1rpo intends to go through them eventually (although not too soon i think due to the many examples, time, motivation etc, but it is to be expected i mean and i would be happy if he goes through them eventually especially if it is in depth i mean so i would expect some time it would take to start it and go through it i mean).
I hope advciv can be made stronger/more interesting i mean to play against which is my main motivation for it xd, for me, possibly for others too i mean, so if it can help all good i mean anyways,

As for debugging and logging and such i would leave it to you guys i mean if you can follow-up, these tools would add another layer of detailing which may be a bit too much so i'd rather focus on reporting the things i think i think are important or/and that i am curious about and i think are significant i mean anyways.
I hope that the way i structured my examples by numbers in the same download folder, with an explanation for each example in the specific post would help keep things organized, also as @f1rpo i try to provide the save file a turn before the decision happens or closest to it, rarely i provide a range of save files when the behaviour is a set i mean of many decisons that happen over a length of time i mean (for example example 1 and no scientific method research at all for dozens of turns, or example 16 and the back and forth of the stack i mean anyways).

thanks for giving me motivation and feedback too, if it's too much i'd rather not do it xd, so as i am playing i find it easiest to just submit a save file with a new example increment of each new situation i see, i hope that renaming the save file is also easier to navigate through ingame while switching save files too i mean anyways.
 
I have another weird behaviour i noticed i mean.
It is example 28.

The incan settler is 2 tiles south of Antioch (my city i mean), in the same time as my axeman (that i used to scout to investigate where the settler went).

Maybe i just don't see the tile where the settler is the next turn, and maybe i am mistaken, but unless i am mistaken, an incan settler mysteriously i mean vanishes.
It is escorted by 2 archers i mean, yet somehow the archers retreat without a settler with them, and the settler is mia xd (unless i am mistaken i mean).

So the first problem in this example is why go there if not to settle? inefficient, so maybe either go somewhere else or don't create a settler or create a settler but don't go there just to go back, the navigation of the settler is inefficient i mean, why navigate all this just to go back.

The second and main problem is unless i'm mistaken didn't this settler just disappear? How is it even possible. I tried to scout with the axeman south next turn but i don't see any unguarded settler nor a settler escorted by another unit. Also north and west i don't see the settler in any of ragnar's tiles. So where did the incan settler go i mean. How can the escort be alive i mean but the settler gone.
I don't know how to use the cheats you mentionned i mean and for now would not want to spoil myself, but i mean do you have an explanation about this.

Secondly i mean, what do you think of the inefficiency of the settler moves. Is this normal/intended, would you want to improve it in advciv, for example with the other alternatives i discussed/proposed in this post i mean. Thanks,
 
Back
Top Bottom