Realism Invictus

[[Y]]

Well, we're basically running back into essentialist nature of Civ 4 civilizations, especially as embodied by RI. We all know that Civ bonuses only vaguely make sense in real world context, so on random maps they should theoretically get completely different ones (inland Austronesians, or tundra Mayans...).

Roger that. Reflecting on it, I think this might be part of my bitterness (or stubbornness, depending on one's perspective ). I think RI does a good job with this overall, where distinct units will have a moderate advantage in their historical terrain (such as celtic units receiving a 10-15% bonus in hills), and national units having a strong relationship with historical terrain (such as celtic Gaestate starting with Guerilla I and Fianna starting with Woodsman I and II). Distinctive units are slight variations on a theme (for better or worse) and replace the standard unit, and national units represent a big difference in strategy, but exist in addition to the standard roster, so they aren't replacing anything. The distinctive skirmishers are a departure from this with some having national unit level bonus replacing a standard roster unit. It would be more consistent if the bonus difference between desert skirmishers and forest skirmishers amounted to a plus or minus of 15%, rather than the full 50%.

One additional gameplay consideration about forests is that they are easily removable, with one of the first in-game techs, unlike almost everything else terrain-related. So having or not having a forest next to your city is a tactical choice (whereas, a desert, for instance, goes nowhere the whole game). That doesn't invalidate what I said above about hiking indefinitely in enemy territory but clearing forests in the immediate city radius is always an option (and likely the reason for such a high defensive bonus in the first place).

True, but it takes 25 worker turns to remove a forest. If a city has 3 adjacent forsets, that's 75 worker turns. More if there are more forests and even more when you have to do this for each city. There are also resources like Prime Timber, Deer, Fur, and Pigs where you might want to leave the forest for a production bonus, though that's an intentional tradeoff, so maybe not as big of a deal. Still though, if a civ has 3 cities and needs to remove 10 forest tiles (250 worker turns) before an enemy civ shows up with a stack of units (as the AI likes to do), that's a lot of investment, and means investing in removing forests rather than building other improvements, road connections, etc. I also think that this doesn't have much do to with some civs having a massive forest bonus on skirmishers and others not.

That would still need to create basically the same AI framework to process - currently, there is no assumption at all made by AI that terrain/features can provide different protection depending on ownership.

Ahh, okay. My assumption was that the AI looks at a tile and sees how much total defense a unit would have there, and doesn't have an understanding of how that value is arrived at in the first place (so no concept for hill/forest in its calculations). But if it's actually looking at the features themselves and drawing conclusions, yeah, I see how that's a major hurdle.

Yeah, for me it's just a throwback to the early design process of the tech tree - "where do we put triremes?"; the more all-encompassing nature of the tech probably warrants Weapon Smithing to be a prerequisite to Trireme unit class specifically rather than the tech itself. Looks like the early tech tree is due for another minor shakeup.

For full transparency, the question came from a very selfish place. The Great Lighthouse is one of my go-to wonders. The AI tends to get to it pretty quickly, and easily since it's just a follow up to Weapon Smithing, while I usually aim for some econ/religion techs before working up to it. So my want was to access the econ wonder more easily by investing in econ techs rather than having to detour through Weapon Smithing first.

 

[Walter Hawkwood]

Ahh, okay. My assumption was that the AI looks at a tile and sees how much total defense a unit would have there, and doesn't have an understanding of how that value is arrived at in the first place (so no concept for hill/forest in its calculations). But if it's actually looking at the features themselves and drawing conclusions, yeah, I see how that's a major hurdle.

It is part that, but partly an even higher level of decision-making: for instance, if AI doesn't have enough units to attack and awaits reinforcements, does it fall back to its own territory or does it fortify where it is? Currently, there is no assumption that the ability for a stack to protect itself would be fundamentally different in either case. Does an AI need a routine to actively lure enemies to its own territory to have an advantage if we introduce such an advantage? Stuff like that.

For full transparency, the question came from a very selfish place. The Great Lighthouse is one of my go-to wonders. The AI tends to get to it pretty quickly, and easily since it's just a follow up to Weapon Smithing, while I usually aim for some econ/religion techs before working up to it. So my want was to access the econ wonder more easily by investing in econ techs rather than having to detour through Weapon Smithing first.

I'll see what feels natural. I did have a feeling that the trade branch of the tech tree wasn't integrated enough with the rest of the techs...

 

[Ahnarras]

I'll see what feels natural. I did have a feeling that the trade branch of the tech tree wasn't integrated enough with the rest of the techs...

Yeah, I had the same feeling about early stone age tree, where you could usually go pretty far in middle row / bottom row without having to bother with top row.
And as those tended to have most of the military & wonders, top row was usually the last that were worked on, except if I was drowning in marine ressources or in a need for galleys.

Making Mathematics (thus the whole top row, with Mathematics needing Trade) a needed tech for urban planning solved that.

I can see how having something similar for Weapon Smithing could help. Still, it also needs to make sense, and by looking at the Tech Tree, I can't really finds a Tech that makes me go "oh yeah, you obviously need that to smithe weapons".
There's Metal Casting that could works, but it's a bit later than Weapon Smithing.
Perhaps Mathematics, again ? Or would that put too much weight on that particular Tech ?
There's also Alphabet, but I'm not sure how having an Alphabet would be mandatory for smithing weapons...

I also thought about making Weapon Smithing unlock later, to put it more in line with Metal Casting, but then you have S5 Axeman unlocking at the same time as S6 Composite Bowmen. A bit weird, the Weapon Smithing is in a good spot as a "middle of the way" tech.

 

[[Y]]

Yeah, I had the same feeling about early stone age tree, where you could usually go pretty far in middle row / bottom row without having to bother with top row.
And as those tended to have most of the military & wonders, top row was usually the last that were worked on, except if I was drowning in marine ressources or in a need for galleys.

Making Mathematics (thus the whole top row, with Mathematics needing Trade) a needed tech for urban planning solved that.

I can see how having something similar for Weapon Smithing could help. Still, it also needs to make sense, and by looking at the Tech Tree, I can't really finds a Tech that makes me go "oh yeah, you obviously need that to smithe weapons".
There's Metal Casting that could works, but it's a bit later than Weapon Smithing.
Perhaps Mathematics, again ? Or would that put too much weight on that particular Tech ?
There's also Alphabet, but I'm not sure how having an Alphabet would be mandatory for smithing weapons...

I also thought about making Weapon Smithing unlock later, to put it more in line with Metal Casting, but then you have S5 Axeman unlocking at the same time as S6 Composite Bowmen. A bit weird, the Weapon Smithing is in a good spot as a "middle of the way" tech.

Yeah, I agree with this. It's why I supplemented the idea with moving units/buildings around, so that the techs can stay as is, but access to things gets more balanced.

 

[PKSRoman]

Its weird that Cruel / Conqueror /Imperialistic or Militaristic so rarely decide to change to a dictatorship - more often I've seen them prefer monarchies or change from monarchy to democracy in peacetime and to monarchy when they declare war on someone

 

[Takofloppa]

I'm making quite good progress with fixing revolutions - there were a few key issues I identified that made AI struggle with controlling it (overusing spy specialists, bad scaling of unhappiness/unhealthiness modifiers through the game, slow responding to high separatism levels), so hopefully my changes will land in SVN somewhere in the near future. I still have to do a few autoplays and balance some things, because the line between toothless revolutions and the world in fire is thin

 

[PKSRoman]

Seondok Silla turns Autocracy while she have option to turn into Dictatorship (She`s Top1),revolts risde and city turn into barbarian rule

 

[[Y]]

Just noticed that the Mughal empire has turkic units--Should it be the Moghal empire instead? Seems like the Moghul and the Mughal got mixed up.

 

[globosud]

In my opinion too much is expected of A.I. of civ 4, too much complexity cannot be managed, the paths of the a.i. They must be forced, the A.I. is elementary, it does not know how to make the best choice in relation to certain conditions, if, for example, a leader is militaristic he must follow an 80% obligatory path, civilizations cannot have specific promotions linked to land, because, the a.i. will never be in a position to exploit them to his own advantage. Similar choices are fine for a scenario, where you know the starting positions of all the civilizations, where you already know where resources or other things are located.

 

[[Y]]

Just noticed that the Mughal empire has turkic units--Should it be the Moghal empire instead? Seems like the Moghul and the Mughal got mixed up.

Nevermind this, I just wasn't well read enough about the two and their linked history. Though also goes to show how complicated history gets. The Mughals were founded by a descendant of Timur, a turk, who in RI is a Transoxianan leader, who's units ancient/classic era units are Median and Bactrian, not turk. I guess the lineage is represented in RI with the Mughals having a mix of turkish and hindi units, rather than identifying as one or the other.

In my opinion too much is expected of A.I. of civ 4, too much complexity cannot be managed, the paths of the a.i. They must be forced, the A.I. is elementary, it does not know how to make the best choice in relation to certain conditions, if, for example, a leader is militaristic he must follow an 80% obligatory path, civilizations cannot have specific promotions linked to land, because, the a.i. will never be in a position to exploit them to his own advantage. Similar choices are fine for a scenario, where you know the starting positions of all the civilizations, where you already know where resources or other things are located.

That's a good point, and I often forget to consider AI when proposing ideas, so I'm glad you brought it up. But there are probably ways to accomplish the same with less player agency, allowing the AI to benefit from it as well. Though a few months back Walter added a change allowing barbs units to promote according to the nearby terrain (or something along those lines), so that aspect of AI decision making isn't implausible.

 

[globosud]

Nevermind this, I just wasn't well read enough about the two and their linked history. Though also goes to show how complicated history gets. The Mughals were founded by a descendant of Timur, a turk, who in RI is a Transoxianan leader, who's units ancient/classic era units are Median and Bactrian, not turk. I guess the lineage is represented in RI with the Mughals having a mix of turkish and hindi units, rather than identifying as one or the other.


That's a good point, and I often forget to consider AI when proposing ideas, so I'm glad you brought it up. But there are probably ways to accomplish the same with less player agency, allowing the AI to benefit from it as well. Though a few months back Walter added a change allowing barbs units to promote according to the nearby terrain (or something along those lines), so that aspect of AI decision making isn't implausible.

let's say that the barbarian unit does not need strategy, because it only has one objective: to destroy us, while civilizations, in theory, could have multiple objectives, for example saving the army, or giving up the attack, therefore basically, the path of the barbarians is quite predictable, for example giving a promotion that can also be used for human players, could bring the AI. to crash without hope, surrounded by a well-positioned army. To make a change that works really well, you should calculate all the possible variables, but there will be thousands. Then if you also add the change of leader, then it becomes almost impossible to calculate all the variables. then it also depends on what the player is looking for, whether one wants a truly competitive challenge or just wants to enjoy multiple contents. in my experience, all the mods that advertise modifications of this type or even more complex are simply broken, and the A.I. ends up being very easy to beat. the only way the A.I. have to win is brute force, with many free bonuses.

 

[JDCP]

the only way the A.I. have to win is brute force, with many free bonuses.

And that is sadly why Noble, allegedly the most balanced difficulty setting, is not really fair at all. The player is given way too many liberties regarding what he can do, whereas the AI as you said is set on a path. The only way to sort of ''fix'' this is to play in the hardest difficulty which always feels like a kick in the balls, and even then you still got the advantage.

One issue with RI is that AI is definitely not designed to take in account most changes in combat, or at least fails at it, so you will see stacks that go over the logistics limit (which is not necessarily bad, but the AI takes this to a extent where it can become trouble) or using units that definitely aren't made for whatever purpose they use them. I've used stacks of 18 Levies and that's alright agaisn't the computer, but agaisn't a player I would probably have to resort to a different strategy than just brute force because many weaknesses come with that, weaknesses the AI does not know how to exploit. Also they tend to use whatever units they got around for a siege rather than specialized shock troops and cavalry (your generic assault divisions), so it's not uncommon to see stacks with archers and pikemen (not that they are bad for that purpose, but without the right promotions, aids and some support they suck).

This issue is minimal though, I enjoy it the way it is because mastery of the game or the AI representing a big trouble for me have never been something of my likeness.

That's the reason multiplayer is the only real way to have a fair challenge, until then, AI will only beat you if you allow it to do so. AI's have evolved dramatically in the last years but I still don't see them winning over us in a CIV game lol.

the path of the barbarians is quite predictable

And keeping the AI predictable is, we like it or not, a good thing mostly. It gives you control over things and takes randomness out of the equation. You don't want to play a game where your enemy is a maniac who does whatever a RNG tells him to.

I've seen other mods who try to modify the AI to a extent where it ruins the flow of the game completely or makes diplomacy impossible, I know what you mean

 

[Takofloppa]

I would love to improve AI's understanding of logistics and stack bonuses one day, it's just that there isn't a straightforward way to do it (otherwise it would have been done a long time ago) with the current implementation which in the worst case can require total rework, or at least I have no idea at this moment. Fortunately, it isn't that much visible from the player's perspective and the lack of AI's stack quality is usually compensated by quantity.

 

[Ahnarras]

Yeah, R:I seems to have reduced the "stack of doom" mentality even for the AI, at least it's the impression it gives me.
I still see "big" stacks from time to time, like 15-20 units triggering the logistic penalty, but it seems to be always compensate by the huge number of stack aid bonuses thoses tiles enjoy.

 

[Walter Hawkwood]

Its weird that Cruel / Conqueror /Imperialistic or Militaristic so rarely decide to change to a dictatorship - more often I've seen them prefer monarchies or change from monarchy to democracy in peacetime and to monarchy when they declare war on someone

Traits don't directly influence civic choices. All AI leaders generally have their own preferences that weigh in on their civics (though not overwhelmingly so, they still "think" practically about their choices). But one thing other thing (@Takofloppa, for your notes I guess, as you're the one working on separatism AI) is that separatism reduction is currently, IIRC, not factored into AI civic choices in any way at all.

I'm making quite good progress with fixing revolutions - there were a few key issues I identified that made AI struggle with controlling it (overusing spy specialists, bad scaling of unhappiness/unhealthiness modifiers through the game, slow responding to high separatism levels), so hopefully my changes will land in SVN somewhere in the near future. I still have to do a few autoplays and balance some things, because the line between toothless revolutions and the world in fire is thin

Good! I'm considering releasing 3.71c soon-ish, as I'll probably make some unrelated AI changes myself. When testing 3.71b I accidentally got myself a perfect set of saves of an AI leader making an idiotic war declaration decision that 100% shouldn't be happening - so I'm going to take that situation apart and likely put it back together with some war AI changes. My preliminary assessment tells me there might be no "bug" to fix - it's just that K-Mod rewrote this part from scratch, and the new logic might be, by design, allowing some really stupid decisions, such as attacking someone who's twice as strong as you, under certain circumstances.

Nevermind this, I just wasn't well read enough about the two and their linked history. Though also goes to show how complicated history gets. The Mughals were founded by a descendant of Timur, a turk, who in RI is a Transoxianan leader, who's units ancient/classic era units are Median and Bactrian, not turk. I guess the lineage is represented in RI with the Mughals having a mix of turkish and hindi units, rather than identifying as one or the other.

Thanks for not making me write a wall of text on Mughal history! I really enjoy it when RI prompts people to learn something new about history. Mughal founder Babur was quite a character! He actually wrote an autobiography, which is a rather entertaining read for a historical source, as it details the extent of his (mis)adventures before actually founding one of the greatest contemporary empires. That's a guy who got kicked down ten times and sprung back eleven...

Though a few months back Walter added a change allowing barbs units to promote according to the nearby terrain (or something along those lines), so that aspect of AI decision making isn't implausible.

Not quite. What I did was add an AI logic that severely restricted its ability to pick terrain-dependent promos for its units, and in a rather hacky way. It is a really daunting task to think of a piece of code that would really decide in any detail whether an AI unit needs a desert promo without a lot of additional calculations that would have a severe negative performance impact.

And keeping the AI predictable is, we like it or not, a good thing mostly. It gives you control over things and takes randomness out of the equation. You don't want to play a game where your enemy is a maniac who does whatever a RNG tells him to.

I've seen other mods who try to modify the AI to a extent where it ruins the flow of the game completely or makes diplomacy impossible, I know what you mean

Yeah, I feel that people often misunderstand the general purpose behind the existence of AI in games. It is not there, in general, to win against you, or rather, trying to win is just part of its "job description". The whole purpose of it being there is to entertain you, and while being a credible threat is a major component of that (though how major varies greatly from player to player), but even more importantly, it shouldn't be annoying while doing that. Making someone ragequit is the opposite of entertaining them.

Yeah, R:I seems to have reduced the "stack of doom" mentality even for the AI, at least it's the impression it gives me.
I still see "big" stacks from time to time, like 15-20 units triggering the logistic penalty, but it seems to be always compensate by the huge number of stack aid bonuses thoses tiles enjoy.

RI is generally geared mechanically towards requiring fewer units, not just from AI standpoint, but across the board.

 

[RamsesII]

Hi all,

A few gameplay questions (am on my second game, Egypt, large earth map, Monarch difficulty, year us 184AD). Still on R:I v3.7.

- unlike the Bantus, the Swahilis do not seem to have tribal forts (unless they have disbanded them early on?), which makes them an easy target for the Zulus. Is this intentional?

- the Israelis build their UB the fortress, which replaces walls. The pedia says it provides -75% to city bombardment (vs -50% for normal walls). But when I bombarded with my battering rams, it did nothing. Is it a rounding thing, or has the fortress been coded to provide -100% to city bombardment? as a consequence Jerusalem has been besieged for centuries by almost all its neighbours, the land around is pillaged and barren, but the archers behind the city's fortress hold it and it seems virtually impossible to take it...

 

[giorgio1234]

Hi,
is there a way to save the map options of the RI_Planet_Generator script?
There is a saveMapOptionDefaults method in the script, that should do what I ask, but unfortunately I can't find the saved file anywhere, and the settings are not retained from one game to the next.
Can anyone help me?

 

[f1rpo]

There might be an error message in PythonDbg.log (provided that LoggingEnabled is set in My Games\Beyond the Sword\CivilizationIV.ini). When I just tried with my own copy of the script (the RI version looks the same to me as far as the cfg file is concerned), the file was successfully created in and read from My Games\Sid Meier's Civilization 4, i.e. the base game directory. I've just pushed a fix for that (moving it to the BtS dir) and a few other issues to my GitHub repo: link
But, if the config file wasn't working at all before, then this probably won't help either. Or it might in the case of a BtS standalone edition(?).

 

[Walter Hawkwood]

- unlike the Bantus, the Swahilis do not seem to have tribal forts (unless they have disbanded them early on?), which makes them an easy target for the Zulus. Is this intentional?

Yes, Zulus need at least a bit of Lebensraum.

- the Israelis build their UB the fortress, which replaces walls. The pedia says it provides -75% to city bombardment (vs -50% for normal walls). But when I bombarded with my battering rams, it did nothing. Is it a rounding thing, or has the fortress been coded to provide -100% to city bombardment? as a consequence Jerusalem has been besieged for centuries by almost all its neighbours, the land around is pillaged and barren, but the archers behind the city's fortress hold it and it seems virtually impossible to take it...

I thought I fixed that one ages ago, but apparently I didn't. Thanks for reporting!

There might be an error message in PythonDbg.log (provided that LoggingEnabled is set in My Games\Beyond the Sword\CivilizationIV.ini). When I just tried with my own copy of the script (the RI version looks the same to me as far as the cfg file is concerned), the file was successfully created in and read from My Games\Sid Meier's Civilization 4, i.e. the base game directory. I've just pushed a fix for that (moving it to the BtS dir) and a few other issues to my GitHub repo: link
But, if the config file wasn't working at all before, then this probably won't help either. Or it might in the case of a BtS standalone edition(?).

Thanks, I suspect that was the culprit indeed. Since I have the full version, I wouldn't catch the fact that it doesn't find the base Civ 4 folder. I think that was it; @giorgio1234 are you an SVN user? I could push the change for you to test.

On an unrelated note: IIRC, all you need for the map yield symbols to work is indeed the contents of Art/Interface/Symbols, and setting MAX_YIELD_STACK in GlobalDefines to 20 (obviously unreachable under vanilla game rules, but it shouldn't hurt anything, and the current files are built with 20 in mind). Also, please feel free to ask if you need any assistance with any other art assets, or even some custom stuff - I will be delighted to help you out with anything.

 

[giorgio1234]

Hi,
is there a way to save the map options of the RI_Planet_Generator script?
There is a saveMapOptionDefaults method in the script, that should do what I ask, but unfortunately I can't find the saved file anywhere, and the settings are not retained from one game to the next.
Can anyone help me?

ok, I understood where the problem was.
The script tries to create the file in the folder "C:\Users\User\Documents\My Games\Sid Meier's Civilization 4", that doesn't exist on my system (as I have never launched civ4 vanilla, but always and only RI ).