Buffed AI for BNW

[glider1]

Well I've had 20 downloads not much to worry about distracting people from improving civ! Neither this tiny mod nor Thal's mod stop work on the tactical AI.

Let's see, I've got about 1 hour per day that I can spend on playing/modding civ. That gives me just enough time to tweak this mod, play the game and take notes of where I think the AI could be best improved and most quickly. That's about all I can do!

I don't get holidays over Christmas/New Year. Last holiday I had was two years ago. Sometimes, I get a bit more time and look at the game code more deeply but unfortunately the outlook appears to be less time for civ not more (but who knows) and I could well be heading for the breadline in 2014 (who knows).

I help where I can to give people alternatives and to at least keep me playing this game in some way shape or other. Anything to keep the campfire lit.

 

[Aristos]

Well I've had 20 downloads not much to worry about distracting people from improving civ! Neither this tiny mod nor Thal's mod stop work on the tactical AI.

Let's see, I've got about 1 hour per day that I can spend on playing/modding civ. That gives me just enough time to tweak this mod, play the game and take notes of where I think the AI could be best improved and most quickly. That's about all I can do!

I don't get holidays over Christmas/New Year. Last holiday I had was two years ago. Sometimes, I get a bit more time and look at the game code more deeply but unfortunately the outlook appears to be less time for civ not more (but who knows) and I could well be heading for the breadline in 2014 (who knows).

I help where I can to give people alternatives and to at least keep me playing this game in some way shape or other. Anything to keep the campfire lit.

Any help for the better AI mod being born will be welcome and important... you know C++, right? Can I add you to the team?

 

[Marshall Thomas]

Hey, glider1. I just started a new campaign with this mod and I'm enjoying it.

The only thing I see in the notes that could potentially worry me is that the AI receives zero happiness bonus. Ideally, this would be great; but can the AI handle it? Is this compensated for in some other way?

Thanks in advance and thanks again for creating this mod. For me, it's essential.

 

[Darkwave09]

Been playing the new version of this mod a lot lately on diety. The start of the game is great but later on the AI can't keep up scientifically. Around the industrial era they start to fall behind. By the modern they are behind. Atomic even further, by the info era I'm 8-15 techs ahead of them. 20-30 techs ahead of the smaller weaker AI. By the start of the industrial era they no longer DOW. The improvements you made with version 8 (I think) work great at the start of the game. Many DOW from the AI with large armies. Sometimes I win, sometimes I loose. Could you boost the AI at the later eras? This mod works great.

 

[layelaye]

Hi, could you maybe make a mod that buffs the science per era a lot more? right now the diety AI cant keep up with me at all.

if the AI can win a science victory around turns 250-270 it will be a match for the best players out there, and all that's necessary is a bigger science buff per era!

 

[glider1]

Thanks for the report Layelaye good to hear that the other aspects of this mod seem to adjusted ok except for the late era science.

Did you notice any carpet of doom with the deity AI? Recently when I study the unmodified deity AI, towards the end of the game they have ridiculous amounts of units (carpets of them) because they can build them so cheaply and don't have to wear the same maintenance cost. Preventing that carpet of doom on deity is another thing this mod should be achieving, because it is not giving the deity AI the ability to build units for next to nothing.

 

[Marshall Thomas]

Thanks for the report Layelaye good to hear that the other aspects of this mod seem to adjusted ok except for the late era science.

Did you notice any carpet of doom with the deity AI? Recently when I study the unmodified deity AI, towards the end of the game they have ridiculous amounts of units (carpets of them) because they can build them so cheaply and don't have to wear the same maintenance cost. Preventing that carpet of doom on deity is another thing this mod should be achieving, because it is not giving the deity AI the ability to build units for next to nothing.

Glider1, I just wanted to say that I've really been enjoying this mod on emperor. To me, it's so nice to be challenged without having to play on immortal or deity.

 

[glider1]

That's great to hear Marshall, how much should I boost per era science and should I merge this with Smart AI mod?

Estimations could go something like this I guess:

Science boost per era (on top of what the AI does now)
Ancient 0%
Medieval 0%
Renaissance 20%
Industrial 50%
Modern 75%
Information 100%

The problem is this mod is supposed to make it challenging without forcing the game into who can fill the science bucket the quickest!

 

[Marshall Thomas]

That's great to hear Marshall, how much should I boost per era science and should I merge this with Smart AI mod?

Estimations could go something like this I guess:

Science boost per era (on top of what the AI does now)
Ancient 0%
Medieval 0%
Renaissance 20%
Industrial 50%
Modern 75%
Information 100%

The problem is this mod is supposed to make it challenging without forcing the game into who can fill the science bucket the quickest!

I would definitely like to see this merged with Smart AI mod. In fact, I was about to request it.

Personally, I don't think science-per-era needs to be boosted from where it is now; for the exact reason you mentioned -"this mod is supposed to make it challenging without forcing the game into who can fill the science bucket the quickest." Maybe there could be an option to customize the science-per-era percentage. Personally, I like it as it is.

also: I'm trying to create a very small mod which only does one thing: gives all AI units +1 to movement, like Persia's units during golden ages. I noticed the Persian AI maneuvering considerably better (especially in counter-attacks while I was sieging their cities with artillery). This made me think that this could be something to "level the playing field" between the player and the tactical combat AI; in the same spirit as the +33% attack bonus for the AI in your mod.

Could you please take a look at these two threads?
http://forums.civfanatics.com/showthread.php?t=527326

http://forums.civfanatics.com/showthread.php?t=527368

I'd really like to hear your thoughts on this. Thanks in advance.

 

[glider1]

I'll submit a new version for the weekend, with Smart AI and customizable per era science which will allow us to fine tune it.

I've tried +1 movement a year or so ago in my own experiments giving it to the AI but not the human, because I too thought it would help the AI. What I found is that the AI breaks it's formation and it plays ugly. For example, one melee unit will skip ahead of the rest if it calculates it has the movement points, arrive at the enemy front line on its own and get slaughtered before the rest of the army catch up. It all depends on terrain movement opportunity. If the map were a chess board, I think +1 movement could work, but it is not. So I dropped the idea. However, perhaps it is still worth considering.

 

[Marshall Thomas]

I'll submit a new version for the weekend, with Smart AI and customizable per era science which will allow us to fine tune it.

Thanks! I'll never play another game of CiV without it.

I've tried +1 movement a year or so ago in my own experiments giving it to the AI but not the human, because I too thought it would help the AI. What I found is that the AI breaks it's formation and it plays ugly. For example, one melee unit will skip ahead of the rest if it calculates it has the movement points, arrive at the enemy front line on its own and get slaughtered before the rest of the army catch up. It all depends on terrain movement opportunity. If the map were a chess board, I think +1 movement could work, but it is not. So I dropped the idea. However, perhaps it is still worth considering.

Interesting. It seems like we're thinking along the same lines (considering similar solutions for making CiV challenging without having to play on immortal or deity).

I wonder if Smart AI, with it's fix to AI move and shoot problems, could possibly make +1 movement for the AI better than it was when you tested it a while back.

Your +33% damage for AI attacks is doing a really good job of making the player lose units more often. Now even successful wars come with some cost; and military catastrophes are possible now. They were completely impossible before.

One issue that keeps coming back to me in all my games is artillery always saving the day and providing the turning point. No matter how much I tell myself that I want to break my strategic cycle of always beelining dynamite and then finally start warring when I can attack cities safely from 3 hexes away... it always plays out that this is simply my best move to make. So naturally I wonder about it's balance.

I use to play immortal/deity, but with Buffed AI, emperor is challenging (that's the main reason I like it so much). So I'm usually in a very close, competitive situation and the dynamite/artillery strategy is almost always what saves me. That's part of the reason I was thinking of the +1 movement for the AI: so that it could make more aggressive counter-attacks to break through to my artillery when I'm beginning a siege.

Thanks again for this mod and merging it with Smart AI mod. I haven't tried your More War mod yet, but I'll think I'll do so soon. Cheers!

 

[glider1]

Yeah, the main benefit of smart-AI is that ranged know how to move and shoot which will really hurt in this mod. I have a doubt that smart-AI will help the AI understand how to use extra movement points better.

Anyway, agree, beelining artillery all the time is very ugly. If I remember, siege are considered ranged, and so human bombard units should already be getting nerfed by the NERF_HUMAN_RANGED value of -25%. What about if I make it so that human bombard units only get an opportunity attack against AI units? So their actual offensive attack against AI units is in the random range of 0% to 75% but against cities it is always 75%. That makes some historical sense because artillery are better at attacking a large target, not individually dispersed infantry. That should make Dynamite less attractive to us humans and will cost us more money because siege won't be as useful and if the human brings in more siege to compensate, they will be more difficult to protect.

On another topic, have a theory why Darkwave thinks the AI science rate is too low late game and you think it is not. If players play this mod on archipelago type maps where the AI's are significantly separated from humans, the extra declare war probability of this mod is probably not going to kick in because the AI will think it is silly to declare war on a distant human. If the AI is not threatening militarily, the human is not being handicapped as much by this mod, because they don't have to spend as much on military.

So science progress in this mod is ESPECIALLY sensitive to AI crowding around the human.

What settings do you play on Marshall 8+16 on land maps?

 

[Marshall Thomas]

Anyway, agree, beelining artillery all the time is very ugly. If I remember, siege are considered ranged, and so human bombard units should already be getting nerfed by the NERF_HUMAN_RANGED value of -25%. What about if I make it so that human bombard units only get an opportunity attack against AI units? So their actual offensive attack against AI units is in the random range of 0% to 75% but against cities it is always 75%. That makes some historical sense because artillery are better at attacking a large target, not individually dispersed infantry. That should make Dynamite less attractive to us humans and will cost us more money because siege won't be as useful and if the human brings in more siege to compensate, they will be more difficult to protect.

I like the idea of nerfing Artillery (exactly how is a difficult question), but not cannons. Going from 2 to 3 range is just such a quantum leap! -not to mention indirect fire. I thought about maybe lowering artillery's range to 2, but compensate by giving it greater attack strength. However I eventually dismissed this idea because the industrial and post-industrial battlefields are (and rightly so) more crowded; so needing to be within two hexes to start shelling a city would leave too few frontline troops in front of your siege units. You would end up with excess melee units behind your artillery (three hexes away from the city -instead of in the second line, behind the front line).

I don't really think that cannons should be nerfed any more. Napoleonic/Fredrick the Great's cannons were used on the battlefield to fire on troops directly. First World War artillery (and later artillery) was used to fire on fixed positions from far away; and though they were often very inaccurate, they could deal lots of damage without even needing to be in a battle (which is why they killed so many -often on a daily basis). In CiV, I think this represents shelling a city more than troops. So part of the problem is that, despite the fact that they seem to be over-powered in some ways, CiV does a decent job in representing the difference between Napoleon's cannons and First World War artillery. Having said this, I'm still looking to nerf them in order to break my cycle of strategic dependency. I have read many other posts where players note the huge leap from range 2 to 3 and indirect fire... all in one unit upgrade; saying that it's far too great a jump for one unit. Originally, I agreed with this; but, as I said, changing their range or their indirect fire ability is not, imho, the solution.

On another topic, have a theory why Darkwave thinks the AI science rate is too low late game and you think it is not. If players play this mod on archipelago type maps where the AI's are significantly separated from humans, the extra declare war probability of this mod is probably not going to kick in because the AI will think it is silly to declare war on a distant human. If the AI is not threatening militarily, the human is not being handicapped as much by this mod, because they don't have to spend as much on military.

So science progress in this mod is ESPECIALLY sensitive to AI crowding around the human.

What settings do you play on Marshall 8+16 on land maps?

I just upgraded my pc a few months ago and finally had the chance to try playing on 'huge' maps (previously I could just barely play on 'large' maps, with all the graphics settings set to their lowest -with DX9). However I found huge maps to be a little too big for my liking. So many civs and city-states became a chore to keep track of and, if I were to play with an attention to detail (which I like to do), my game would become very slow and somewhat tedious.

So now I'm playing on a modded size that's in between large and huge, but closer to large (about 1/3 the way to huge). Also, I always play on 'continents' maps.

In my current game, I'm playing with 15 civs and 22 city-states. The ratio of 1:2 is too many city-states for my personal taste. With twice as many city-states as civs, I find that the borders between civs are too often blocked by city-states. There were never any spacious fronts like the eastern front of WWII, which really limited wars and, because city-states aren't usually good for conquering, civs seemed too safe behind these buffers of city-states. I was worried that lowering this ratio might unbalance things, but fortunately I haven't found that to be the case.

I always play on epic speed. I tried marathon, but it was a little too extreme for me. A speed in between these two might be nice; maybe slightly closer to epic.

The only other mod I use is a slower research mod. It's simply slows research down by about 15%, I believe (I've been using it forever, so I haven't seen the exact numbers in quite a while).

I play with policy saving enabled, as I don't think you should ever be penalized for having too much culture. I also play with promotion saving enabled, though this seldom comes into play. I made the mistake of trying 'complete kills' in my current game. I won't use it again.

I play with all victory conditions enabled.

I also have one house-rule: I can't try to bribe another civ to attack another unless I'm already at war with the civ I want attacked. I feel like it's already too easy to manipulate the AI as it is. and I can't really imagine Britain (during WWII) desperately trying to get the US to declare war on Germany if they hadn't already been at war with Germany themselves.

Also, as I mentioned earlier, I now play on emperor -thanks to your mod (use to play on immortal). It's much more enjoyable now, while being equally, if not more, challenging.

Thanks in advance.

 

[glider1]

Ok, what about nerfing human artillery indirect fire at range 3 but not against cities? So at range-3 they only do random 0-75% damage on units? Against cities it remains constant at 75%

 

[Marshall Thomas]

Ok, what about nerfing human artillery indirect fire at range 3 but not against cities? So at range-3 they only do random 0-75% damage on units? Against cities it remains constant at 75%

Nice. I had no idea that you could change damage percentages for different ranges and for indirect fire! That opens up several possibilities.

Your suggestion sounds right. This would be good to have customization for, so we could tweak it until we find a good balance.

That's really great news that damages at different ranges and direct/indirect fire is moddable. As far as I know, no units in the vanilla game have different damage at different ranges; so I just figured it wasn't possible.

Thanks again.

 

[glider1]

Almost anything is possible when you work in the DLL like this mod does. How about I just make a flat rule that anything ranged at distance 3 only gets opportunity fire from random 0 up to the NERF_HUMAN_RANGED limit (but against cities the nerf is constant at NERF_HUMAN_RANGED). This means that Englands longbow also get a nerf. The idea for this is basically that humans are so much better than the AI when it comes to range 3 because there are so many more options opened up that the AI does not properly consider (and which would eat into the computation time if it did).

 

[Marshall Thomas]

Almost anything is possible when you work in the DLL like this mod does. How about I just make a flat rule that anything ranged at distance 3 only gets opportunity fire from random 0 up to the NERF_HUMAN_RANGED limit (but against cities the nerf is constant at NERF_HUMAN_RANGED). This means that Englands longbow also get a nerf. The idea for this is basically that humans are so much better than the AI when it comes to range 3 because there are so many more options opened up that the AI does not properly consider (and which would eat into the computation time if it did).

That makes perfect sense to me. This would apply to battleships as well? I think this would be a reasonable nerf to them.

Also: what is opportunity fire? Is it simply offensive fire (damage done from attacks you initiate)?

Thanks in advance.

 

[Marshall Thomas]

sorry. double post

 

[glider1]

Yeah it's a term I have borrowed from our Civ4 mod RevolutionDCM which was part of Dales Combat Mod. So yes it is only offensive and simulates the idea that accuracy drops off with range but it is worth taking an opportunity shot anyway...

I could scale it but the difference between range-2 and 3 is too coarse. I doubt it is a good idea to nerf humans with a bit of less random damage even at range 2. In any case, it is an easy rule to remember that at range 3, offensive damage is lower in a random fashion. Yes it would apply to ranged 3 navy as well.

I've already merged in Smart AI and the coding for this will take about 1/2 an hour.

I've got two questions for you Marshall:

In this mod, the AI get's no bonuses on religion (other than standard production bonuses) so that humans have more strategic variety. Do you think the AI is spreading it's religion at roughly the rate a human could?

Could you post the grid size of the modified large-huge map you use or at least the link to where ever you got it? Thanks.

 

[Marshall Thomas]

Yeah it's a term I have borrowed from our Civ4 mod RevolutionDCM which was part of Dales Combat Mod. So yes it is only offensive and simulates the idea that accuracy drops off with range but it is worth taking an opportunity shot anyway...

I could scale it but the difference between range-2 and 3 is too coarse. I doubt it is a good idea to nerf humans with a bit of less random damage even at range 2. In any case, it is an easy rule to remember that at range 3, offensive damage is lower in a random fashion. Yes it would apply to ranged 3 navy as well.

I've already merged in Smart AI and the coding for this will take about 1/2 an hour.

That's great new about the merger of the two mods. So soon too! Will it be saved game compatible?

I 100% agree that there's no need to scale. Ideally units with a range of 2 are suppose to fire from 2 hexes away; should not be more powerful from one hex away. With the obvious exception of gatling guns and machine guns, you never should have an incentive to have your range units one hex away from the enemy. For units with a range of 3, it makes perfect sense that damage would be less from so far away. and it makes greater sense when we consider just how much better the player is at utilizing range 3 artillery than the AI is. In my current game, I don't think I could have possibly won without artillery and tithing. As I mentioned in an earlier post, artillery has bailed me out of so many games that I feel something has to be done. It's no fluke when it happens every time; but I can't help myself if not using artillery means I won't win.

I've got two questions for you Marshall:

In this mod, the AI get's no bonuses on religion (other than standard production bonuses) so that humans have more strategic variety. Do you think the AI is spreading it's religion at roughly the rate a human could?

I think so. In my current game, I'm about to win a cultural victory. The key was beating China (the cultural/tourism front-runner) in a long, but decisive war and taking their wonders. The reason I mention this with regard to religion (and it's spread) is because I cannot imagine having been successful in this game without having tithing. At one point it was bringing in 90 gold-per-turn. Without it, I couldn't have wrecked China. I don't think it was so much a case of the AI not being sufficient at spreading it's religion (compared to me), as much as the possibility that tithing is maybe overpowered. The AI spread it's religion fairly well; keeping my religion out of entire continents almost completely. That's why I'm thinking that tithing is overpowered: if I had really spread my religion much better than the AI, than it may not be that tithing is over-powered, but rather that the AI is inept at spreading it's religion; but that really wasn't the case. I've considered modding tithing to be one gold for every 3 citizens, instead of every 2.

Having said that, I think that a modest boost to the AI's missionary efforts may be a good idea.

Could you post the grid size of the modified large-huge map you use or at least the link to where ever you got it? Thanks.

My larger, 'large' map:

<Type>WORLDSIZE_LARGE</Type>
<Description>TXT_KEY_WORLD_LARGE</Description>
<Help>TXT_KEY_WORLD_LARGE_HELP</Help>
<DefaultPlayers>15</DefaultPlayers>
<DefaultMinorCivs>21</DefaultMinorCivs>
<GridWidth>111</GridWidth>
<GridHeight>68</GridHeight>
<MaxActiveReligions>6</MaxActiveReligions>
<FogTilesPerBarbarianCamp>30</FogTilesPerBarbarianCamp>
<NumNaturalWonders>7</NumNaturalWonders>
<UnitNameModifier>10</UnitNameModifier>
<TargetNumCities>6</TargetNumCities>
<NumFreeBuildingResources>6</NumFreeBuildingResources>
<BuildingClassPrereqModifier>75</BuildingClassPrereqModifier>
<MaxConscriptModifier>50</MaxConscriptModifier>
<TerrainGrainChange>1</TerrainGrainChange>
<FeatureGrainChange>1</FeatureGrainChange>
<ResearchPercent>120</ResearchPercent>
<NumCitiesUnhappinessPercent>80</NumCitiesUnhappinessPercent>
<NumCitiesPolicyCostMod>7.5</NumCitiesPolicyCostMod>
<NumCitiesTechCostMod>3.75</NumCitiesTechCostMod>
<AdvancedStartPointsMod>110</AdvancedStartPointsMod>
<EstimatedNumCities>80</EstimatedNumCities>
<IconAtlas>WORLDSIZE_ATLAS</IconAtlas>
<PortraitIndex>4</PortraitIndex>
</Row>


Default large:

<Type>WORLDSIZE_LARGE</Type>
<Description>TXT_KEY_WORLD_LARGE</Description>
<Help>TXT_KEY_WORLD_LARGE_HELP</Help>
<DefaultPlayers>10</DefaultPlayers>
<DefaultMinorCivs>20</DefaultMinorCivs>
<GridWidth>104</GridWidth>
<GridHeight>64</GridHeight>
<FogTilesPerBarbarianCamp>30</FogTilesPerBarbarianCamp>
<NumNaturalWonders>6</NumNaturalWonders>
<UnitNameModifier>10</UnitNameModifier>
<TargetNumCities>6</TargetNumCities>
<NumFreeBuildingResources>6</NumFreeBuildingResources>
<BuildingClassPrereqModifier>75</BuildingClassPrereqModifier>
<MaxConscriptModifier>50</MaxConscriptModifier>
<TerrainGrainChange>1</TerrainGrainChange>
<FeatureGrainChange>1</FeatureGrainChange>
<ResearchPercent>120</ResearchPercent>
<NumCitiesUnhappinessPercent>80</NumCitiesUnhappinessPercent>
<NumCitiesPolicyCostMod>10</NumCitiesPolicyCostMod>
<AdvancedStartPointsMod>110</AdvancedStartPointsMod>
<IconAtlas>WORLDSIZE_ATLAS</IconAtlas>
<PortraitIndex>4</PortraitIndex>