3.6 AI-only Game Statistics

[pineappledan]

The biggest question mark to me is whether continuing to increase the min influence of Austria is basically just copying Greece's influence reduction.

Wasn't Austria's old 100% arrested decay even more of an overlap then? At least raising the resting influence is a different mechanism than a flat % decay reduction at all levels.

If you wanted to change Greece, You could lower the CS decay bonus and increase emphasis on the combat bonus. Could lower the CS decay reduction to 25% slower, but raise the max to combat bonus 50% (capped at 10 CS allies, up from 5).
I've always thought Greece's CS combat bonus seemed a bit weak. +25% doesn't go very far after the mid game.

 

[Stalker0]

Wasn't Austria's old 100% arrested decay even more of an overlap then? At least raising the resting influence is a different mechanism than a flat % decay reduction at all levels.

not disagreeing with the recent change we made, but not sure continuing to push that further is also ideal especially as a complete replacement for the bonus votes.

 

[pineappledan]

Making the votes 1 per X marriages makes them more complicated. It makes marriages' value highly variable. It's another thing to keep track of in addition to the 5 turn tracking for marriage. All this added complication to preserve a 3rd component in a single kit that that doesn't unlock until Renaissance (in addition UU and UB), and that we all agree is not even close to balanced. Blegh.

This bonus sucked on Germany, and it sucks here.
If we get rid of the WC vote, the marriages become weak enough that we can make them cheaper and not require the 5 turn waiting period. We can make this ability easier and safer to use as Austria while also being more fun to play against Austria. win-win.

edit:

Adding 1 vote for every 3 marriages won't even physically fit in the UA description, which already fails to mention the 5 turn alliance requirement for marriages because it doesn't have the space.

 

[Solic]

The biggest question mark to me is whether continuing to increase the min influence of Austria is basically just copying Greece's influence reduction.

I do think the votes have to be reigned in; I don't know if they need to be removed entirely. Maybe 1 votes per 3 marriages or something. But the 1 vote per marriage is just too crazy, I agree that Austria is often the civ in the game that you HAVE to destroy in order to win, their power is just too all encompassing

You can also use them to your advantage if you become their biggest friend and make their votes work for you or vassaling them alternatively. Staying alive early as Austria should remain rewarding in my opinion.

I don't understand the argument why UA are special over UU/UB in needing to completely show a payoff early on. Civ kit components work together and some get online later than others. That's a variety I personally enjoy. I can see the argument against having any "late game civ", but that's more up to personal preference, so congress can do its work I suppose.

Note that I agree that Austria requires a nerf given the AI games stats.

 

[Stalker0]

here's a version after a lot of debate on discord, so people can weigh in.


**Habsburg Diplomacy**
+100% rewards from City-State quests. Can marry Allied City-States using Gold, which increases Resting Influence to Influence at the time of the Marriage. Gain +1 Great Diplomat Point and +10% to Great People generation in the Capital for each Marriage with a peaceful City-State.

The idea

1. Remove the vote bonus which is the most OP part of Austria's kit.
2. Return some CS flavor with an increase in the CS quest rewards and a GD GPP point bonus for the marriage.
3. Reduce the GPP % from 15 to 10 on the marriages. This focuses Austria more in her GD kit while reducing her massive GP advantage

 

[Amask]

Disagree about removing the 5 turn delay.
It's really not that complicated to keep track of.
Regardless of whether marriages give votes or not, it serves an important function of providing counterplay.
Other civs should have an opportunity to prevent Austria from marrying everything.
Simply allying for a turn isn't that hard to achieve.
Keeping that alliance when the CS is close to someone else and that someone else wants to spam envoys is quite a bit harder, and it should be.

 

[pineappledan]

I would submit that the 5 turn wait time is only necessary because:
1. The marriages are so powerful right now, and
2. The scaling Cost on later marriages is so low, and isn’t an impediment.

Requiring allied status without stipulating X full turns is still an impediment to allying everyone, but the possible danger of that happening is significantly less without those WC votes. It’s a powerful bonus still, but now so powerful that I feel we need specific counterplay cutouts built into it so that other civs can resist the onslaught.

It’s also a really opaque stipulation. It’s only mentioned inside the CS screen. There is no space in the UI text elsewhere to mention this limitation. It’s not great how this UA hides information like that right now.

Edit: also what could be more Habsburgian than utterly bankrupting yourself trying to stitch together a loose coalition of states into an empire too large, unwieldy, and disloyal to maintain?

 

[der_weihnachtsmann]

For me part of the problem is, that as a player you seem to need a lot more votes to benefit from WC, because often the other two proposal are directly against your interest, so you kind of need more votes. AI often doesnt seem to have this problem.

here's a version after a lot of debate on discord, so people can weigh in.


**Habsburg Diplomacy**
+100% rewards from City-State quests. Can marry Allied City-States using Gold, which increases Resting Influence to Influence at the time of the Marriage. Gain +1 Great Diplomat Point and +10% to Great People generation in the Capital for each Marriage with a peaceful City-State.

The idea

1. Remove the vote bonus which is the most OP part of Austria's kit.
2. Return some CS flavor with an increase in the CS quest rewards and a GD GPP point bonus for the marriage.
3. Reduce the GPP % from 15 to 10 on the marriages. This focuses Austria more in her GD kit while reducing her massive GP advantage

Whenever I play Austria, the most satisfaction I get is pressing the DM button, foremost because I dont need to worry about the current (and less the future) influence. Thats why I am against making the resting influence dependent on current influence while I agree with all the other changes. Additionally, I feel like its very realistic to end up being a buff, because of how fast you can stack influence, although it often means a bigger investment. This might be especially true for AI with a lot more resources, which would also need to be trained to know the best amount of influence to freeze.

Edit: I know considered the 100 % quest rewards and it should actually be a piece of cake to get to more then 200 influence.

 

[pineappledan]

This might be especially true for AI with a lot more resources, which would also need to be trained to know the best amount of influence to freeze.

If we manage to make Cost a legitimate concern for marriages, that curtails players’ ability to both spike a high influence level while still being able to afford the marriage. Diplo marriages will happenat higher levels later in the game as a natural product of more competition between civs for CS allies. Worst case scenario is a civ manages to still increase their resting influence by 60.

Therefore, I don’t know if there would be any training required, since there is no wrong time to marry, just trade offs.

 

[azum4roll]

Can you say it's always best to do the marriage as soon as you ally the CS, and teach the AI to do so? You can't change your mind if you happen to complete a quest one turn later that bumps up the influence by another 120.

 

[azum4roll]

Current influence decay formula for reference:

If influence is lower than resting point:

Influence change per turn = 1 + Same religion bonus (0.5) + Greece UA bonus (1)​

If influence is higher than resting point:

Base change rate B = -2 if the civ has declared war on a CS twice, -1.5 if CS is Hostile, -1 otherwise​

Extra decay E = floor((Current influence - Resting influence) / 100) ^ 1.5​

Fail to protect modifier F = 300% if CS capital is damaged while civ has pledged to protect it, 100% otherwise​

Decay modifier D = 1 - Same religion bonus (25%) - Greece UA bonus (50%)​

Final influence change = (B - E) * F * D​

​

How the rounding down works creates a jump in influence decay, and makes being 1000 influence above resting point worse than 999, for example. The rounding should be done after the exponent operation instead.

 

[CppMaster]

For me part of the problem is, that as a player you seem to need a lot more votes to benefit from WC, because often the other two proposal are directly against your interest, so you kind of need more votes. AI often doesnt seem to have this problem.

So that would mean that AI treats the player differently in that regard, which is an issue. Do devs confirm it? @Recursive

 

[der_weihnachtsmann]

So that would mean that AI treats the player differently in that regard, which is an issue. Do devs confirm it? @Recursive

I think on higher difficulties the AI has a negative opinion modifier towards the player right?

 

[CppMaster]

I think on higher difficulties the AI has a negative opinion modifier towards the player right?

I hope not. There is victory competitiveness enabled by default, but it's a separate thing.

 

[Recursive]

So that would mean that AI treats the player differently in that regard, which is an issue. Do devs confirm it? @Recursive

I think on higher difficulties the AI has a negative opinion modifier towards the player right?

No in both cases. isHuman() isn't even checked in ScoreVoteChoiceYesNo. When it comes to selecting diplomatic approaches, the only difference with humans is that the AI assumes they'll use any nukes they have on them 100% of the time (some AIs can also have this likelihood).

They do have a positive opinion modifier towards the player on Settler and Chieftain.

 

[CppMaster]

No in both cases. isHuman() isn't even checked in ScoreVoteChoiceYesNo. When it comes to selecting diplomatic approaches, the only difference with humans is that the AI assumes they'll use any nukes they have on them 100% of the time (some AIs can also have this likelihood).

Cool! I'm relieved

 

[Recursive]

Spoiler :

OK, but how do you know this data isn't dominated by the AI personalities/preferences (the main point of my post)? I think some of them might be extremely disadvantageous

This is what I'm thinking. There are probably some misrepresenting values along the civs personalities and flavors. Also, some civs' kits have lower flavor than it supposed to have, like Babylon.

Babylon is pretty OP in human hands, but not so much in AI hands. Walls of Babylon flavor value is one of the culprits, it isn't changed at all from the previous version so it doesn't affect AI priority for getting it much sooner.

I'm sure this is also the case for other civs. I would say why not we try to re-adjust the flavor values for all the unique components for civs that are failing to properly use their kit?

This is an unfounded claim. I would need to see some actual analysis comparing game results and correlating them with certain approach or win flavours before I take it seriously.
It will be very hard to make a convincing argument that it is the flavours that are hurting AI and not their victory emphases, because flavours for civs oriented towards certain victory types (eg. a domination civ) tend to be very similar.

The AI flavours are suited to the civs and their kits. Civs oriented towards domination, for instance, have high war and offense flavours. You can be reasonably sure that all AI will perform substantially worse with random flavours.
I am not, however, saying that all the flavours are perfect. I think Morocco, for instance, is too passive for his own good. But if there is an effect on win rate, it is not large.

Aside from that, @Recursive put in a monumental effort to adjust flavours to make them more suitable, and even created a third set of primary and secondary VictoryPursuit attributes that guide AI diplomacy and win biases. These VictoryPursuit bonuses are set individually for each civ and aim them toward 1 or 2 most likely victories, based on their overall kit, and the biases for each civ seem quite reasonable to me.

No, the far more likely case is that the AI is either not particularly good at using a component, irrespective of their flavours, or the component is weak, plain and simple. I think it does an incredible disservice to the years of work that people like Recursive have done on the diplomatic and AI flavors to cast aspersions at them here.

I didn't run the games, but if I had I would have played them on Warlord, where the AI has no bonuses, so the bonuses cannot bias results.

That being said, there is merit to running the tests on Emperor, which is the difficulty that the devs have publicly made statements about trying to balance around.

perfect balance would mean removing all differences between civs; we don't want balance if it comes at the cost of unique abilities etc.

This can't possibly be right.

The flavours may not be as high as you would like, but they are higher than any other building babylon is going to aim for in that era. As I understand it, the AI also doesn't consider building flavors all that much anyways

No, WoB flavor value is not any higher if you compare it with other Science buildings.

WoB and Library have the same FLAVOR_SCIENCE value, also not that far off than Council. It might conflict with AI decision-making. I would suggest bumping the value a little bit, so then Babylon can make the building a priority.

An unfounded claim? My "claim" is that the effect of flavours on winrate is currently unknown, relative to UAs etc. Your claims, that the effects of flavours are not large etc etc, are what is unfounded, since there is essentially no data that could currently support them. I certainly appreciate the effort of making this data and previous efforts to balance the flavours (and everything else). I'm just saying, why not simulate the flavours separately?


If civs are balanced around AI personalities, how does that affect the player?

So you’re proposing that AI games should be run without any attempt to get the AI to play to a civ’s strengths? What would that do? The flavours will still all be turned on in a regular game, so that won’t simulate anything close to a normal game. The AI games reflect a real game environment at the default difficulty, size, speed, and map settings that are the main target to be balanced for. AI players with no flavours is not a default setting.

We have also discussed if the tests could have been run without AI difficulty bonuses. There are a myriad of things that aren’t being controlled for, but calling it “silly”, for someone to spend weeks of their time to run hundreds of games in a way that wasn’t precisely how you would have done it is not cool.

Regardless, it is not the data we have, so there isn’t much sense in discussing it. I’ll defer to recursive, who knows more about the AI personalities, and to Vern, who knows much more about his test environment. If you want to run your own test games with wiped AI personalities I think he could help you get started.

What I'm suggesting is that AI flavours could and should be simulated without civ abilities, and civ abilities without flavours (or with random personalities), if people are serious about balancing using AI games. What would that do? I think it's obvious and I've explained it, but it would let us see how strong the effects of flavours are compared to civ abilities. At a minimum I would also want to see both Warlord and Emperor data to take balancing conclusions seriously. Personally I think balancing using AI games has a lot of other issues, but you seem extremely keen to do it, and I have no doubt you possess the energy to keep on posting until whatever you have in mind comes into being. So really pineapple, could you not yourself put some of that energy towards generating some more meaningful data? After all you're the one who wants to implement a whole list of changes based on this data, but I think I've given some valid criticisms that a number of people here agree with.

If you’re unhappy with the data presented thus far, that doesn’t make it my, or Vern’s or anyone else’s responsibility to satisfy you. You’re the one who seems to have energy to burn and a mouth to run.

Many changes have already been implemented without a dataset this robust, and I would prefer this approach to what we have been doing up to this point. This particular dataset took weeks for Vern to generate and I did, in fact, spend some of my own time giving input on it as it was being made.

So yes, I will continue to advocate for balance changes based on test games. I have very little patience for people saying we have too little, or the wrong kind of data to be making changes. Gazebo used to run AI test games as his main way of balancing, and We have gone years since his departure moving the mod forward without anything but user feedback. We’ve never had this level of transparency or input to interpret and discuss test game results until now, so I’m aghast that this is suddenly not good enough for some people.

As I've said, I am not unhappy with the data, I just think it's insufficient to justify balancing decisions. You say it's robust, but I say it isn't. At least including separate data for flavours would make it so much more meaningful and interesting. I appreciate the effort it takes, I would just really like to see it improved further if it's supposed to be for balancing.

As for "energy to burn and a mouth to run", I really can't think of a better way to describe you. I've seen you doggedly pursue changes to this mod time and time again which are based on your own speculation and superficial aesthetic concerns, rather than actual experience and enjoyment playing the mod. These changes are often implemented because you just keep on going, while ignoring most criticism. I know from experience you can, and will, do this all day and all week. Frankly there's not much I can do other than watch you slowly make VP worse, since unlike you I do not have all day. I can't be bothered arguing with you any more. As usual you haven't actually responded to my points which were presented quite politely and reasonably, and have instead deferred responsibility for your own suggestions and dived into name-calling.

I am quite sure a lot of the results are impacted significantly by flavors and personality, but there are 2 points I think are important to keep in mind as well:
1. The purpose of this exercise is to get an idea of how civs will behave and perform in something close to what people are actually playing on, hence using default size, VP map, settings, most common difficulty

Spoiler :

View attachment 664957

as its under these conditions that most players will be seeing the AI civs and under these conditions I think we should focus on making sure none of them stand out in a not fun or not interesting way (note that in a standard size game there will still be 7 AIs and their interactions with one another will still be better modelled by this approach than with equalizing all of their flavor and personality numbers)
2.
While in general I totally agree that some flavors are disadvantageous and may be a good place to increase performance, having a wide variety of personalities makes the game more interesting and keeping those is preferable in many cases

My relevant work here has been on the flavors for the Diplomacy AI, which have a significant impact on diplomatic behavior and jumpstart the AI's pursuit towards victory conditions. I would argue these flavors are much more significant than any others at the moment.

I have not worked on the following:
- Non-diplomacy leader flavors. These have been given some cursory balancing in LeaderFlavors.sql, but frankly I don't know what most of them do or how they impact the AI's decision-making. From my observations they seem to have a low impact - much of the AI scoring for beliefs, policies, techs, and city production choices is not flavor-based.

- Grand Strategy AI, which decides which victory condition the AI will pursue later in the game if they're competing for victory (as opposed to early in the game or when not competing for victory, where it will go with its chosen victory pursuit). Grand Strategy AI does make use of non-diplomacy leader flavors in ways I'm not sure about. Gazebo rewrote this code a long time ago.

In both of these two areas, I think there's opportunities for improvement.

Regarding running the AI with random personalities, I fail to see how the AI would gain anything from having Gandhi act like Askia or Hiawatha act like Haile; they're adapted to the Diplomacy AI's playstyle. For a detailed guide to what they are and how they work, see this thread: https://forums.civfanatics.com/threads/vox-populi-diplomacy-guides.683886

There are some adjustments based on the leader's historical context but it's heavily focused on making the AI play better. There is also some randomness to make the AI more unpredictable and not "feel" exactly the same from game to game. Testing without any flavors sounds like a waste of time to me, but if you want to anyway, here are the steps you'd need to follow:

Spoiler :

1) Replace the content of LeaderPersonalities.sql with the following lines of code:

UPDATE Defines SET Value = '0' WHERE Name = 'FLAVOR_RANDOMIZATION_RANGE';
UPDATE Leaders SET PrimaryVictoryPursuit = NULL;
UPDATE Leaders SET SecondaryVictoryPursuit = NULL;
UPDATE Leaders SET Personality = NULL;
UPDATE Leaders SET VictoryCompetitiveness = 5;
UPDATE Leaders SET WonderCompetitiveness = 5;
UPDATE Leaders SET MinorCivCompetitiveness = 5;
UPDATE Leaders SET Boldness = 5;
UPDATE Leaders SET DiploBalance = 5;
UPDATE Leaders SET WarmongerHate = 5;
UPDATE Leaders SET DoFWillingness = 5;
UPDATE Leaders SET DenounceWillingness = 5;
UPDATE Leaders SET Loyalty = 5;
UPDATE Leaders SET Forgiveness = 5;
UPDATE Leaders SET Neediness = 5;
UPDATE Leaders SET Meanness = 5;
UPDATE Leader_MajorCivApproachBiases SET Bias = 5;
UPDATE Leader_MinorCivApproachBiases SET Bias = 5;
UPDATE Leader_Flavors SET Flavor = 5;

2) Delete LeaderFlavors.sql in the same folder.

Moderator Action: The namecalling and personal attacks are not appropriate. I see the argument has ceased but there will be moderation action if it resumes. Be civil to each other.
Please read the forum rules: http://forums.civfanatics.com/showthread.php?t=422889

 

[Gazebo]

If you’re unhappy with the data presented thus far, that doesn’t make it my, or Vern’s or anyone else’s responsibility to satisfy you. You’re the one who seems to have energy to burn and a mouth to run.

Many changes have already been implemented without a dataset this robust, and I would prefer this approach to what we have been doing up to this point. This particular dataset took weeks for Vern to generate and I did, in fact, spend some of my own time giving input on it as it was being made.

So yes, I will continue to advocate for balance changes based on test games. I have very little patience for people saying we have too little, or the wrong kind of data to be making changes. Gazebo used to run AI test games as his main way of balancing, and We have gone years since his departure moving the mod forward without anything but user feedback. We’ve never had this level of transparency or input to interpret and discuss test game results until now, so I’m aghast that this is suddenly not good enough for some people.

Has it really been years? Ah jeez...