XML tags for the Planetfall Leaders

[Maniac]

Thanks for the file!

 

[Maniac]

- Domai will now ask you twice as often to stop trading with someone else (he also values such actions more, so it is consequent that he asks more often for them as well)

Do you ever break open borders because an AI asks you to? Especially since IIRC the AI never asks other AIs to break open borders, so I'd rather see this behaviour changed to the same as the other AIs.

I was also thinking about removing the Consensus civic. The "no unhappy for non-state religions" could be moved to Democratic. Seems a more interesting bonus than just some fixed extra happiness in your largest bases. I could then make Planned the favourite civic of the Consciousness, and Wealth of the Drones.

 

[Pfeffersack]

Do you ever break open borders because an AI asks you to? Especially since IIRC the AI never asks other AIs to break open borders, so

I'd rather see this behaviour changed to the same as the other AIs.

Rarely. Being consistent with the philosophy of not over-emphasizing AI-human-only diplomatic issues, I agree (though I still think it would make sense in the context of the drones philosophy).

I was also thinking about removing the Consensus civic. The "no unhappy for non-state religions" could be moved to Democratic. Seems a
more interesting bonus than just some fixed extra happiness in your largest bases. I could then make Planned the favourite civic of the Consciousness, and
Wealth of the Drones.

Given that Consensus seems to be a (another) underused choice anyway and it would make it easier to set the Consciousness and Drones favourite civics, I think that a good idea. Fits also pefectly to Democracy. If we now just get Fundametalism
right so that AIs will consider it at least occasionally as a choice like Police State...
perhaps give Fundi a 3rd free XP point (and take one away from Power if necessary)?
Or +50% longer resistance when a base base get captured...`
Or spread of partisan units when a base gets captured...
Or increased building speed for missionaries...

-----

I think the addition of the expansion factions will make some further modifications necessary. They have without a doubt spiced up the game and brought in a
lot more action - on the other hand, they also catalysed some negative mechanics:

1. The vasall mechanics might need rework outside of the leaderhead.xml. We have two types of vasalls - conquered ones and peaceful ones - and the leaderhead.xml only offers one parameter for each to characterize a faction. For peaceful vassalization, you can set a mood treshold - the future vassall needs to think high enough of a potential vasall. For capitulation, there is just a positive or negative value setting if a leader is rather (un)willing to capitulate. The main problem I see is that the capitulation parameter only fits for isolationlists/realpolitikers like Yang or Morgan - almost never capitulate, just for the sake of staying independent or give in to a stronger to survive. A Deidre or Miriam would probably surrender to someone with similar goals, but fight till death vs. a heathen. And that is something you cannot achivec with that purely strength-based value.
And capitulations like Miri to Zhak(!) in my last game seem to turn over the entire game, because the attitude of teams is somehow middled, so that clear
diplomatic blocks/clevages vanish completly

2. The result of 1) in combination with more potential aggressors leads to more war declarations and faster detoriation of relations because of the -3 for "You declared war on us" (leading to more wars again and so on...). Thats why I think we need to make the AI forget declarations at some point. I have posted a solution in this thread a while ago and I would suggest to implement this with a middle-value for the "speed of forgetting" for everone for the moment. We can still differnentiate the factions later. We might try 50 or 100, which is a common value (ok, the AI needs 150 turns to forget that you refused a demand, but that's IMO too much anyway).
BTW, drastically increasing "forget speed" for AI-human-only-actions generally would be another way to balance them.

3. Maybe we should finetune the war attitude percentages also - I have e.g. Deidre (0/50/100/100) in mind. Now that she directly factors your PA into her feelings, extrem values (like furious/friendly) get more likely anayway. Furious means no protection from war plans anyway (no value here), but the current design is that annoyed is already 0 for Deidre. That means that it does not really matter how much you set her up - I would prefer to change her values to 50/75/100/100, so that she will focus on the Terraformers more
The Cautious value was anyway the one I always had most trouble with - how willing should leaders be to attack someone they have neutral feelings against? Sure, that should count too much for Yang/Santiago (but even they shouldn't destroy all relationships - sometimes they need allies...), but for some characters 100% might be interesting. That still does not mean necessarily that they cannot attack you at cautious relations - they can be still bribed into wars or react to other diplomatic issues with war (they just will not consider you as a target in their strategic war plans)

 

[Maniac]

If we now just get Fundametalism right so that AIs will consider it at least occasionally as a choice like Police State...

I'm planning to let Fundamentalism double the effect of your state religion.

1. The vasall mechanics might need rework outside of the leaderhead.xml

I've put it on the to-do list, but I kinda consider it low priority compared to other stuff I've planned. Another option is, when a faction capitulates, to replace its leader by some generic leader who adopts the conqueror's policies. Makes sense to place some puppet leader in a vassalized faction;

2. The result of 1) in combination with more potential aggressors leads to more war declarations and faster detoriation of relations because of the -3 for "You declared war on us" (leading to more wars again and so on...).

May I ask around what turn AIs start to declare wars in your games? I play with Aggressive AI, yet no one ever declares war until 150-200, at which point I quit because I get bored with the lack of action. I want someone to declare war on me! :doitnow!: I don't know. Perhaps I'm just not waiting long enough. I'm kinda getting worried though that there is again some problem in the code which prevents the AI from declaring war.

3. Maybe we should finetune the war attitude percentages also - I have e.g. Deidre (0/50/100/100) in mind.

May I first ask what these numbers mean?

 

[Pfeffersack]

May I ask around what turn AIs start to declare wars in your games? I play with Aggressive AI, yet no one ever declares war until 150-200, at which point I quit because I get bored with the lack of action. I want someone to declare war on me! :doitnow!: I don't know. Perhaps I'm just not waiting long enough. I'm kinda getting worried though that there is again some problem in the code which prevents the AI from declaring war.

That's interesting. I already stopped playing PF with Aggressive AI a while ago (after one of my major adjustments to the file and deeper understanding of the AI war declaration logic, I realized that it was just no longer needed), so my experience with too many wars is without AAI Of course difficulty might have an impact as always...

I browsed the notes for my version k) game and indeed, the first war was in turn 176 the AI declaring on me. I can't say that I'm missing action though...on Immortal I'm busy with keeping up with the expansion speed of the AI, preparing for the inevitable later declaration later and keeping the mindworms away...or close

It is possible that PF has some elements which block earlier declarations - maybe the AI has just enough other "troubles" to deal with. Have you examined which strategies they runned during the peaceful period? This could give an hint on what goes wrong perhaps.

May I first ask what these numbers mean?

I refer to a quadruple of values which comes into the play at the very end of the AI-war-decison-logic. If an AI considers a certain type of war and has passed all other checks with it (so it thinks that it is prepared technologically and numerically and has picked an opponent), then it will a roll a final dice, which might lead to terminating that war plan because of the relation to the victim. Each leader has 4 such values - for leaders he is Annoyed, Cautious, Pleased or Friendly with. Friendly and even Pleased in most cases terminate that plan with 100%, while for annoyed and cautious you have a % chance. It just has the effect that war declarations get more or less often, because more or less of the planning attempts get terminated (furious means that there is no such roll and war will be planned, if the previous checks shave been passed).

 

[Maniac]

Of course difficulty might have an impact as always...

Yeah, perhaps I should increase my difficulty level to Immortal too. It's just that at higher difficulty levels you get the feeling the only way the AI can compete is by massive cheating. That doesn't make for a fun game either IMO.

One other possible difference: do you play with or without Scattered Landing Pods these days?

Have you examined which strategies they runned during the peaceful period? This could give an hint on what goes wrong perhaps.

In my latest game Miriam was planning war on Lal, Lal was planning war on Domai, and Domai was planning war on Lal. They all had to go overseas to reach their target, but they all had transports plus escorts. So I don't know what kept them from declaring. The only thing I can think of is that because they were all planning war, they were all building up their military at the same time, and none of them ever felt they had gained a sufficiently large power advantage to actually declare.

I refer to a quadruple of values which comes into the play at the very end of the AI-war-decison-logic.

Thanks for the explanation! Currently though Deirdre hardly ever declares war at all in my games, so I'd first like to figure out why that is before I halve her chance to declare on people she's Annoyed/Cautious with.

 

[Pfeffersack]

Yeah, perhaps I should increase my difficulty level to Immortal too. It's just that at higher difficulty levels you get the feeling the only way the AI can compete is by massive cheating. That doesn't make for a fun game either IMO.

True, but after a couple of Emperor games played out the same way (me taking the techlead within the first half of the game, then no more competition or threat from the AI), I see this as the minor evil. Immortal AI are at least sometimes dangerous enough to bring me in trouble in regard to surving and will usually prevent me from winning - that's the setting I search for to have fun.

One other possible difference: do you play with or without Scattered Landing Pods these days?

After your hint a while ago, always without.

In my latest game Miriam was planning war on Lal, Lal was planning war on Domai, and Domai was planning war on Lal. They all had to go overseas to reach their target, but they all had transports plus escorts. So I don't know what kept them from declaring. The only thing I can think of is that because they were all planning war, they were all building up their military at the same time, and none of them ever felt they had gained a sufficiently large power advantage to actually declare.

I could really imagine that this has influenced things. Many AI's are very power-sensitive.

Thanks for the explanation! Currently though Deirdre hardly ever declares war at all in my games, so I'd first like to figure out why that is before I halve her chance to declare on people she's Annoyed/Cautious with.

That's another thing I cannot confirm - I was always one of her preferred targets in the last games. And if manage to have peace for a couple of turns, she happily went for someone lese. Also, she has never trouble to fill her bases with native life units. I assume that is similar in your games?

---

When I do the changes to Domai, should I add then the forget code for wars at least? I doubt that this will delay/call of any initial hostilities, but it might increase changes a tiny bit that things can settle sometimes again later...?

 

[Maniac]

True, but after a couple of Emperor games played out the same way (me taking the techlead within the first half of the game, then no more competition or threat from the AI), I see this as the minor evil. Immortal AI are at least sometimes dangerous enough to bring me in trouble in regard to surving and will usually prevent me from winning - that's the setting I search for to have fun.

Okay, I'll try Immortal in my next game. Having looked at the Handicap file, the only thing which worries me comparing those difficulty levels is "iNoTechTradeModifier", which goes from 40 to 30. Do you know what that means? Personally I like t if AIs who are pissed off at you become more dangerous on higher difficulty levels, but I don't like it if increasing your difficulty levels also makes AIs who would usually be friendly towards you also become more hostile - it ruins the potential fun of diplomacy. So I don't know - what does iNoTechTradeModifier do? Does it make everyone less likely to trade techs with you or so.

Also, she has never trouble to fill her bases with native life units. I assume that is similar in your games?

Yeah. In my experience Deirdre is really good at playing Planetfall. In my last game she was in fact the only one more powerful than me, and expanded like crazy.

When I do the changes to Domai, should I add then the forget code for wars at least? I doubt that this will delay/call of any initial hostilities, but it might increase changes a tiny bit that things can settle sometimes again later...?

Actually I'd prefer those values to stay the same. I consider warfare becoming more likely as the game progresses a good thing! My end goal after all is to encourage for instance big climactic wars between hybrids and terraformers to finish off the game in a satisfying manner.

 

[Pfeffersack]

Okay, I'll try Immortal in my next game. Having looked at the Handicap file, the only thing which worries me comparing those difficulty levels is "iNoTechTradeModifier", which goes from 40 to 30. Do you know what that means? Personally I like t if AIs who are pissed off at you become more dangerous on higher difficulty levels, but I don't like it if increasing your difficulty levels also makes AIs who would usually be friendly towards you also become more hostile - it ruins the potential fun of diplomacy. So I don't know - what does iNoTechTradeModifier do? Does it make everyone less likely to trade techs with you or so.

From the Civfanatics Modwiki:

iNoTechTradeModifier Sets a percent modifier for an AI, that determines how many techs a rival can get by trade, before the AI thinks he is becoming too advanced. (100 for chieftain, 70 for noble, 20 for deity)

It would mean that after a AI has traded say 15 techs with you or so, it would stop just because - though there might be factors which reduce this counter over time again.
Shouldn't become a huge problem though - I'm not sure if I have ever triggered this denial reason in PF, since I have switched to Immortal.

Yeah. In my experience Deirdre is really good at playing Planetfall. In my last game she was in fact the only one more powerful than me, and expanded like crazy.

Yes, she absolutely knows how to expand (which is good as that goes well together with Hybrid). If she has problem (beside not attacking in your game), it is using her units effectively. The AI has well-known troubles to understand being the attacker in PSI-combat. Now even that has less consequences for Deidre - she has at least a beefed up PSI-defense- but she still misses out the full advantage she could have.

Actually I'd prefer those values to stay the same. I consider warfare becoming more likely as the game progresses a good thing! My end goal after all is to encourage for instance big climactic wars between hybrids and terraformers to finish off the game in a satisfying manner.

I understand where you come from, but the problem in my last game was that all those war declarations ruined the relations between most players and that seemed to work against those Hybrid/Terraformer climax. Several players had mutual relation well under -30 and the "war coalitions" were usually master and vassal vs. another master/vassal. A third side joined the fun seemingly in a random manner...it was rather chaotic. But maybe I was just unlucky with that game. Or it was somehow influenced by playing LT's Map of Planet (which I really like)

 

[Maniac]

Yes, she absolutely knows how to expand (which is good as that goes well together with Hybrid). If she has problem (beside not attacking in your game), it is using her units effectively. The AI has well-known troubles to understand being the attacker in PSI-combat. Now even that has less consequences for Deidre - she has at least a beefed up PSI-defense- but she still misses out the full advantage she could have.

I need to see the AI in action to see what it's doing wrong and how to improve it. As I've never fought a war with Deirdre, that kinda makes it hard. What exactly is she doing wrong? Her trait ensures her psi defense is the same as her psi attack. And psi units are just as likely as any other unit to use terrain defense boni to their advantage when planning their attack route.

Several players had mutual relation well under -30 and the "war coalitions" were usually master and vassal vs. another master/vassal.

I don't understand. I've never seen such low relations, not in Planetfall, not in FfH, not in vanilla. To get so low, an AI would almost have to immediately redeclare war ten turns on the same player every time the peace treaty expired. Is that the case?

 

[Pfeffersack]

I need to see the AI in action to see what it's doing wrong and how to improve it. As I've never fought a war with Deirdre, that kinda makes it hard. What exactly is she doing wrong? Her trait ensures her psi defense is the same as her psi attack. And psi units are just as likely as any other unit to use terrain defense boni to their advantage when planning their attack route.

To get the 3:2 advantage in land-PSI-combat, you need to be the one to attack. That's not different for Deidre (she just has the advantage that failing in that doesn't give her 2:3, but 3:3 odds - better, but still not 3:2)...and here I see a problem with the AI. A human player will patiently wait (if possible) for the enemy to do the second last move to attack then - the AI is just programmed to march forward and to attack - but it gets attacked in this process very often.

I don't understand. I've never seen such low relations, not in Planetfall, not in FfH, not in vanilla. To get so low, an AI would almost have to immediately redeclare war ten turns on the same player every time the peace treaty expired. Is that the case?

Not -30 from war declarations alone, but together with other stuff. I have seen e.g. -9 from three declarations in my last game. What I don't like about it, is that it is quite huge compared to other things and also absolutely eternal. For atrocities like razing a base or using a nuke...yes, such stuff shouldn't be forgotten easily...but just having declared a war mattering after 300 turns?

 

[Maniac]

To get the 3:2 advantage in land-PSI-combat, you need to be the one to attack. That's not different for Deidre (she just has the advantage that failing in that doesn't give her 2:3, but 3:3 odds - better, but still not 3:2)...and here I see a problem with the AI. A human player will patiently wait (if possible) for the enemy to do the second last move to attack then - the AI is just programmed to march forward and to attack - but it gets attacked in this process very often.

Ah yes true. Unfortunately I have little hope something could be found which addresses this in a satisfying manner. Or perhaps something can be found (I'm remembering some potentially useful function in the SDK now), but it would probably slow down the game significantly.

Not -30 from war declarations alone, but together with other stuff. I have seen e.g. -9 from three declarations in my last game. What I don't like about it, is that it is quite huge compared to other things and also absolutely eternal. For atrocities like razing a base or using a nuke...yes, such stuff shouldn't be forgotten easily...but just having declared a war mattering after 300 turns?

In this particular case it doesn't matter though. -30 and -21 both mean Furious. Anyway, hard for me to say more about this without experiencing more wars myself.

 

[Panopticon]

Ah yes true. Unfortunately I have little hope something could be found which addresses this in a satisfying manner. Or perhaps something can be found (I'm remembering some potentially useful function in the SDK now), but it would probably slow down the game significantly.

That fix would also be likely to lead to many, many stalemates in Psi Combat, given everything else equal. If two AI Psi units see each other, neither will take the first move. If the human meets an AI, s/he will eventually learn that the AI plays this way, and also not take the first step.

Ironically, the bias to the defender in almost all Civ IV combat is what makes large-scale invasions possible, because you can defend your stacks.

 

[Pfeffersack]

Here is a slightly modified leaderhead file (Wealth and Planned as new favorite civics for Domai and Aki and normalized reactions for Domai on Trade Embargos)

 

[Maniac]

Thanks!

 

[Pfeffersack]

Here is the modified HandicapInfo.xml related to the discussion about AIs hating you just for being a human player from Noble difficulty on: http://forums.civfanatics.com/showthread.php?t=338471&page=9 (post #326-)

I kept the boni for Settlers and Chieftains, but for every other difficulty the attitude modifier it is now 0.

 

[Maniac]

Thanks!

 

[Pfeffersack]

I found two interesting threads dealing with the creation of extremly ruthless runaway AIs:

http://forums.civfanatics.com/showthread.php?t=338781

http://forums.civfanatics.com/showthread.php?t=311889&page=1

Looking at the XML for the Runaway AI:

iNoTechTradeThreshold = 50
iTechTradeKnownPercent = 0
TechRefuseAttitudeThreshold = ATTITUDE_FURIOUS​

Be very willing to trade and broker techs. Plainly all AIs should trade techs more. We should also improve the sophistication of the tech trade code.

FLAVOR_GROWTH​

This is to try to leverage the IMPERIALIST trait as much as possible. All AIs with IMPERIALIST should produce settlers more often.

iRefuseToTalkWarThreshold = 150​

If this is a boost, then that means that all the diplomacy that the AI does while it is at war is a net disadvantage, which is unfortunate. We need to make the AI want more in exchange for peace, and have a better idea of what peace is worth to it.

iMaxWarNearbyPowerRatio = 70
iMaxWarDistantPowerRatio = 30
iMaxWarMinAdjacentLandPercent = 0
NoWarAttitudeProbs = 10 for all attitudes​

These and other changes set the Runaway AI's war strategy: it finds a civ that it has a significant power advantage over and it attacks them, regardless of attitude. This is a great war strategy, and all AIs should do it to a greater or lesser extent. If Ghandi sees another civ sitting on 25% of Indian power, and it's not in a vassalage/defensive pact, then that civ is a sitting duck for the warmongers out there. Ghandi should offer a vassage agreement. If the victim civ refuses, Ghandi should attack and capitulate the civ for its own protection.

iSameReligionAttitude* = 0
iDifferentReligionAttitude* = 0
iFavoriteCivicAttitude* = 0​

The existing AI's current religion diplomacy hurts it, so disabling it improves the AI's performance. The same applies for civics. Here's my suggestion for improving religious diplomacy:
http://forums.civfanatics.com/showthread.php?t=337714

UnitAIWeightModifiers - UNITAI_COLLATERAL = +10​

AIs should build slightly more collateral units.

ImprovementWeightModifier - IMPROVEMENT_TOWN = +20​

AIs should slightly increase their valuation of towns.
Edit: on reflection, this probably only helps because of the bug jdog fixed where AI workers keep bulldozing towns.

iBuildUnitProb = 80
iWonderConstructRand = 0​

AIs should build slightly more units and slightly fewer wonders.

Though I don't think the whole thing fits for any PF leader (maybe except the Aliens), I found the read interesting because it shows that even pretty extreme values (can)work. Especially those thinhgs could be fun, if changed for some leaders:

iRefuseToTalkWarThreshold = 150
150 is probably too much, but I somehow feel now that 15 as maximum value in PF is abit low, taking in account that one problem of the AI is that they tend to sign peace, even when they are winning

iMaxWarNearbyPowerRatio = 70
iMaxWarDistantPowerRatio = 30
iMaxWarMinAdjacentLandPercent = 0
Could fit for the rational warmongers. Values below 100 mean that an AI only attacks, if it feel superior. 75 ist currently the lowest for PF and most AIS are 100 or above.
It would probably lead to less wars from those leaders, but the ones started would turn out more effective.

UnitAIWeightModifiers - UNITAI_COLLATERAL = +10
This probably hurts for no one. CD is power.

iBuildUnitProb = 80
iWonderConstructRand = 0
For unit building, 30 is maximum for PF and the leaders just differ in increments of 5.
I think that's just too little difference, gieven that even values as 80 and 90 seem to work. I wouldn't go so far, but 50 or 60 for some leaders could spice up things.
For wonders, it seems that 0 does not mean no wonders (AI will still go for them, if the have rational reasons) - it just means no additional "addiction" for wonder building.
So we could try some leaders with 0 as well.

 

[Maniac]

iRefuseToTalkWarThreshold = 150

For the reason mentioned in the post you quoted, I don't think changing this value is a good approach to creating a better AI.

iMaxWarNearbyPowerRatio = 70
iMaxWarDistantPowerRatio = 30
iMaxWarMinAdjacentLandPercent = 0

I don't like this because
1) I think there are too few wars before turn 200.
2) after turn 200 I'd prefer to see conflicts between Hybrids and Terraformers. Making AIs only declare war at large power differences would run counter to that goal.

iBuildUnitProb = 80

Unit construction is stopped when an AI spends a certain percentage of its income on unit maintenance, so I can see why at a certain point extreme values don't make a difference. You could set unit construction probability at 200 and it wouldn't make a difference. Perhaps a higher value can be experimented with for Yang?

 

[Pfeffersack]

For the reason mentioned in the post you quoted, I don't think changing this value is a good approach to creating a better AI.

I don't really get the argument made there and believe the poster has a wrong idea of what this value does... Naturally, an AI at war with someone does not do diplomacy with this enemy - except peace negotiations. It does diplomacy with third parties still, even more - it will ask them to help in the war or to stop trading with the enemy. The tag in question does only one thing - it prolongues the initial pase after a declarartion where the AI will not talk with the enemy about peace. It does not affect third party diplomacy.

So the question is if it is beneficial for the AI to refuse talking about peace longer. I tend to say yes. I always have the feeling that they give in too fast if they have sucess. Of course the point can be made that this also means they give in later if they loose, but as far as I know a really bad course of the war can override that-no-talk-period.

I think an increased value could help everyone (though there should be still differences - warmongers and ideologic leaders should have a higher value then rest)

Of course the quoted poster is right that changing the AI logic for valuing peave offers might tackle the problem better.

I don't like this because
1) I think there are too few wars before turn 200.
2) after turn 200 I'd prefer to see conflicts between Hybrids and Terraformers. Making AIs only declare war at large power differences would run counter to that goal.

Indeed, 1) might suffer from this. For 2) - especially when combined with greater differences in unitbuildprobabilty - I don't think so, as I would only change some of the leaders (those which care least about that struggle)

Unit construction is stopped when an AI spends a certain percentage of its income on unit maintenance, so I can see why at a certain point extreme values don't make a difference. You could set unit construction probability at 200 and it wouldn't make a difference. Perhaps a higher value can be experimented with for Yang?

I will do some testing then...