Community Ideas: Diplomacy Flavors for AI Leaders

Discussion in 'Community Patch Project' started by Recursive, Jan 25, 2021.

  1. KittyKat3120

    KittyKat3120 Warlord

    Joined:
    Oct 13, 2018
    Messages:
    112
    May I ask are the new diplo flavours added to the recent March beta?
     
  2. Recursive

    Recursive Covets Lands That You Currently Own Moderator

    Joined:
    Dec 19, 2017
    Messages:
    2,691
    Gender:
    Male
    Not the last 10. Others, yes.
     
  3. Hokiefan00

    Hokiefan00 Chieftain Supporter

    Joined:
    Jul 11, 2001
    Messages:
    27
    Gender:
    Male
    Location:
    Weston, Massachusetts
    Would it be useful to share the spreadsheet I've made combining all the flavors? If not, I won't be offended. :)

    I know this is something to add personality to the civs but I figured it could help with making sure one category isn't over/under done.

    I couldn't fit it all in a screenshot and have it legible but here's a snippet:

    Spoiler Civ Flavors :

    upload_2021-3-4_20-6-22.png
     
    gwennog, Ranslee, Hinin and 3 others like this.
  4. Recursive

    Recursive Covets Lands That You Currently Own Moderator

    Joined:
    Dec 19, 2017
    Messages:
    2,691
    Gender:
    Male
    Absolutely, please do! :)

    I was thinking a revision like this would be a good idea, but you beat me to the punch with making a spreadsheet.
     
    Hokiefan00 likes this.
  5. Hokiefan00

    Hokiefan00 Chieftain Supporter

    Joined:
    Jul 11, 2001
    Messages:
    27
    Gender:
    Male
    Location:
    Weston, Massachusetts
    Great!

    Here's a link to the Excel file. Right now it's set to allow anyone to edit it. As I am a looooong time lurker but new participator, let me know if you would prefer it in a different format (Google sheets, etc.) or another storage location other than OneDrive.

    https://1drv.ms/x/s!Av9mteRcCFEPvw8WjFXtypTy4iQt?e=OxbXw9
     
    Chaste, Hinin, LarryAR and 1 other person like this.
  6. pineappledan

    pineappledan Deity

    Joined:
    Aug 9, 2017
    Messages:
    6,714
    Gender:
    Male
    Location:
    Alberta, Canada
  7. Recursive

    Recursive Covets Lands That You Currently Own Moderator

    Joined:
    Dec 19, 2017
    Messages:
    2,691
    Gender:
    Male
    I saved a copy; do you mind if I use this in a new thread (with credit of course :) )? I'll be going over the full list and might tweak a few values relative to each other.
     
  8. Hokiefan00

    Hokiefan00 Chieftain Supporter

    Joined:
    Jul 11, 2001
    Messages:
    27
    Gender:
    Male
    Location:
    Weston, Massachusetts
    I don’t mind at all.
     
    Recursive likes this.
  9. pineappledan

    pineappledan Deity

    Joined:
    Aug 9, 2017
    Messages:
    6,714
    Gender:
    Male
    Location:
    Alberta, Canada
    Looking through @Hokiefan00 for trends that maybe @Recursive and others can assess the skew on some of these, and if it is worth adjusting some leaders in order to give better personality spreads.

    Afraid - heavily skewed down. Max value is 7. Do we want to shift any leaders to be a little more "fearful"?
    Victory Competitiveness - heavily skewed up. Median is 8. Min is 3. 7 are at 10, 1 is pinned at 12. Do we really want 8/43 civs maxed like this?
    Boldness - heavily skewed up. Median is 7. 4 are at 10, 2 are pinned at 12.

    There are no 0 or 11 civs. Is there any sense in making it so some civs have a small chance of being slightly less than totally maxed on a trait?

    Would it be okay if I added a 2nd sheet onto your personality sheet for planning custom civ personalities?
     
  10. Recursive

    Recursive Covets Lands That You Currently Own Moderator

    Joined:
    Dec 19, 2017
    Messages:
    2,691
    Gender:
    Male
    Hmm, so I was afraid (heh) that making leaders too Afraid would lead to the AI being uncompetitive. Few easily scared people became leaders of civilizations, after all. Afraid is the hardest approach to properly weight in the approach calculation because it makes little sense to apply it in most situations, and yet the AI should be Afraid when the situation calls for it.

    Victory Competitiveness is intentionally skewed upwards due to how the flavor is used, and again for competitiveness reasons. Note that the modifier increases with era, so even 4 Victory Competitiveness AIs should be more aggressive by the endgame. Civs with early bonuses cannot afford to delay too long and thus have higher scores here; sometimes I added +/- 1 to account for the leader's personality.

    Boldness could probably use some revision. I'm also going to drop the Warmonger Hate personality element from the combat bonus as it is unpopular and restricts diplomatic options, and make sure those values are fairly distributed.

    Perhaps some 11s or 0s can be added, yeah, although the ingame range will still be 1 to 10. Reviewing it, I think in the last bunch some values were skewed too high relative to earlier civs, so I was going to do some revision anyway. :)
     
    LifeOfBrian likes this.
  11. InkAxis

    InkAxis King

    Joined:
    Feb 24, 2020
    Messages:
    652
    Gender:
    Male
    I feel like looking at the averages like this overlooks some things.
    Not every value has to be perfectly equally distributed.

    Like victory competition, recursive originally said that you should keep this value fairly high. So a skew is good.
    Being afraid shouldn't be a high priority either.
    Boldness can be turned down if it makes sense - but we shouldn't turn stuff down just for the sake of it.

    Edit: seems me and recursive posted at the exact same time :lol:
     
  12. pineappledan

    pineappledan Deity

    Joined:
    Aug 9, 2017
    Messages:
    6,714
    Gender:
    Male
    Location:
    Alberta, Canada
    You know this code better than I do, so you know what is fair or reasonable more than I do.

    However, how I interpret your response is essentially that there are certain personality metrics where the lower bound is simply unusable. If Victory competitiveness' lowest value is Isabella at 3... then maybe Victory competitiveness' AI values should start higher, and Isabella should be made a 1 on that new scale? Functionally, you have a lot of personality traits that only have 8 levels, not 10. Afraid maxes out at 7, so even random chance can't push Arabia or Venice further than 9.

    It's probably not reasonable to do the work of reassigning the values so that the personalities scale more evenly, but it is weird to have a system with so many dead zones.
     
  13. Recursive

    Recursive Covets Lands That You Currently Own Moderator

    Joined:
    Dec 19, 2017
    Messages:
    2,691
    Gender:
    Male
    I think Afraid is okay as-is on the basis that the approach weights are largely relative to each other, not compared to other leaders, and there isn't much functional sense in having leaders of civilizations that are easily frightened. If a modder for whatever reason wants civs to be easier to push around, they can adjust the values themselves.

    Isabella is at 8 Victory Competitiveness, not 3, but yes, you make a good point. It'd take some work to modify this scale, though, and it was easiest in the code to just scale victory penalties to be (penalty modifier * Competitiveness / 10) like the other competitiveness modifiers and skew the value upwards.

    It's really only those two, though?
     
  14. pineappledan

    pineappledan Deity

    Joined:
    Aug 9, 2017
    Messages:
    6,714
    Gender:
    Male
    Location:
    Alberta, Canada
    Hokiefan did wonderful little distribution chartson his excel page, so you can see the gaps yourself. Maybe you can see something I don't notice:
    War mins at 2
    Hostile mins at 3
    Afraid maxes at 7
    friendly mins at 2
    neutral mins at 2
    boldness mins at 2
    denounce mins at 2
    DOF mins at 2
    neediness mins at 2
    Mean mins at 2
    Ignore CS maxes at 8
    Protect CS maxes at 2
    And from that overview alone, you can tell that all values lean high. There are only 4/23 values that have mean/median values of 5 or less
     
  15. pineappledan

    pineappledan Deity

    Joined:
    Aug 9, 2017
    Messages:
    6,714
    Gender:
    Male
    Location:
    Alberta, Canada
    A lot of players are complaining that the AI is gobbling up all the city-states for seemingly no reason, and I can see that pretty clearly reflected in these AI traits.

    There are some civs that benefit more than others from conquering city-states (Aztecs, Assyria, France)
    There are 4 civs that have 10 for CS conquer (aztecs, persia, sweden, Rome). Of those 4, only Aztecs makes much sense to me. I see a weak argument for Rome to be high, but Sweden and Persia, who benefit very little from swallowing city-states, this seems like a distraction.
    For civs with very high Bully traits (Shaka, Mongolia), you might consider making their conquer bias especially low, so they don't get in their own way. Zulu has 12/8 bully conquer and Mongols have 10/6. If you want them to heavily favor bullying, it seems they should not also favor eliminating targets.
    Your ignore traits accross the board are very low. Only 7 civs have an ignore bias higher than 5. This might also be contributing to civs gobbling up CSs so much, because they are all biased towards some sort of action.

    I know it's a bit late to be offering opinions, but Ramkamhaeng's major approach flavors seem off to me. Of the CS-focused civs, he has the most direct combat bonuses from city-state allies. I don't see much wrong with his war approach, but his secondary traits seem at odds with both Siamese history and with his general playstyle, which is to goad civs into war with him with a united front on city-state allies.
    Spoiler Ram flavors :

    Flavour type -- current/recommended
    War -- 5/5
    Hostile -- 4/8
    Deceptive -- 10/10
    Guarded -- 6/6
    Afraid -- 3/7
    Friendly -- 8/5
    Neutral -- 7/4
     
    Last edited: Mar 7, 2021
  16. Hokiefan00

    Hokiefan00 Chieftain Supporter

    Joined:
    Jul 11, 2001
    Messages:
    27
    Gender:
    Male
    Location:
    Weston, Massachusetts
    Go for it. I do enjoy your custom civs!
     
    Hinin likes this.
  17. Recursive

    Recursive Covets Lands That You Currently Own Moderator

    Joined:
    Dec 19, 2017
    Messages:
    2,691
    Gender:
    Male
    Thank you for the feedback.

    I've slept on it and I think it shouldn't be too hard to adjust the VictoryCompetitiveness flavor so that the effective 5-10 scale becomes a 1-10 scale like the others. The additional granularity would be beneficial.

    In the approach calculation, the AI's Conquest weight towards City-States is intended to be zeroed out if there's a major civ they're at war with, but looking at it I've seen a loophole wherein the AI could indeed get distracted, so I'll fix that. I take note of your comments on the ignore/conquest approaches as well, but I'm still thinking about how to address that. Expect some modifications.

    I like your suggestion re: Siam, although I'd tweak it slightly.
     
  18. pineappledan

    pineappledan Deity

    Joined:
    Aug 9, 2017
    Messages:
    6,714
    Gender:
    Male
    Location:
    Alberta, Canada
    question: What would happen if a civ had a very low wonder flavour, but a very high wonder competitiveness flavor?
     
  19. Recursive

    Recursive Covets Lands That You Currently Own Moderator

    Joined:
    Dec 19, 2017
    Messages:
    2,691
    Gender:
    Male
    Wouldn't build many wonders, but would be very mad if they were beaten to them, and would be more likely to conquer civs that were spamming wonders.
     
    KaoticKanine likes this.
  20. Stalker0

    Stalker0 Baller Magnus

    Joined:
    Dec 31, 2005
    Messages:
    8,488
    So the Celts in my game "strongly dislike warmongers". In my head I also thought that the Celts would be more okay with general warmongering, but would be interested to hear the history heads on the forum tell me the real story.
     

Share This Page