1. We have added a Gift Upgrades feature that allows you to gift an account upgrade to another member, just in time for the holiday season. You can see the gift option when going to the Account Upgrades screen, or on any user profile screen.
    Dismiss Notice

Going for Gold: Follower Beliefs

Discussion in 'General Balance' started by Stalker0, May 17, 2018.

?

Is this item in a reasonable state of balance?

Poll closed May 31, 2018.
  1. Yes

    66.7%
  2. No

    33.3%
  1. Workerspam

    Workerspam Prince

    Joined:
    Aug 22, 2016
    Messages:
    324
    I've seen AI go for buildings with their first follower belief more frequently in my games. Per-follower still has an edge but it's no longer a huge gap.

    In past patches I usually had the option to wait on a building (other than Orders, Stupas, and sometimes Cathedrals) if I wanted to take Inspiration or Mastery as my first belief. Not anymore. Orders and Cathedrals remain most popular but I've seen Synagogues, Mosques, and Mandirs often grabbed with the first belief as well.
     
  2. Gazebo

    Gazebo Lord of the Community Patch Supporter

    Joined:
    Sep 26, 2010
    Messages:
    17,848
    Gender:
    Male
    Location:
    Little Rock
    These are just notes, I haven't made any changes yet. As I said I changed the tables so that I could do them in %s instead of 'per follower' or per 2 followers.' The granularity is what I was looking at/for. Really what I was looking for was a consensus on concrete numbers (i.e. too high or too low per one), so thanks to those of you who gave me that info.

    Yes, they are - I've got pretty clear evidence of per-yield followers doing better, primarily because the AI can have a hard time spending its faith in optimal orders if they're expanding or warring. I've been making small tweaks to the formula, but its still a bit problematic.

    Goodness, I don't think I can post any information anymore without it resulting in passionate outbursts. This wasn't intended to be an argument, it was intended to be a 'give me the tweaks you want to see.'

    G
     
  3. Stalker0

    Stalker0 Baller Magnus

    Joined:
    Dec 31, 2005
    Messages:
    6,253
    Its because we are at the end game. So the idea that some of these changes could be the "true and final number to stand the test of time", gets people passionate. But arguments aren't bad as long as they are constructive, and so far I think the argument in this thread is a reasonable one.

    Perhaps we should rethink per follower vs buildings as short term vs investment.

    The religious buildings take a lot of investment, both in faith but really more in time. Its takes a lot of time to get those buildings up. This is doubly so if you plan to reform your religion. Active Spreading has a small and powerful window in the game right now, you really can't afford to go all buildings in your empire, and then switch to missionaries. But that point inquisitors are coming into play and you have to rely on passive spread. But the building bonuses can be quite good and many have strong late game effects

    So what if we allowed the per pop to be focused on early immediate yields. More 1 per 1 pop (maybe per 2) but with small caps. These beliefs could you a more immediate boost, but long term aren't as powerful as the buildings. And then maybe some buildings would need to be switched into a longer term model if they are too "short term" right now.
     
  4. Deadstarre

    Deadstarre Expert

    Joined:
    Nov 1, 2015
    Messages:
    960
    Location:
    New York
    Belief_MaxYieldModifierPerFollowerHalved already existed to use %'s, the base yields provided by Belief_MaxYieldPerFollowerHalved will still exist in some form to function as they currently do? these are useful modding tables for the CP.
     
  5. ElliotS

    ElliotS Warmonger

    Joined:
    Jun 13, 2013
    Messages:
    2,704
    Gender:
    Male
    Location:
    Chicago
    Agree 100%.

    I mean this is clearly the crux of the problem. Many of the per-follower beliefs are either balanced or weak right now, but great for the AI because the AI's religion game is really bad. It's also good for the AIs in human games because they get bonus population, but you've said you test without AI bonuses to avoid that as a problem. (Though Reducing or eliminating growth bonuses from AI could help in that manner as OwlBeBach has done.)

    I think that this will be hard to balance because the AI has such a hard time with religion. In this case I think you'll need to balance around players and then play with the AI's religious code until they get better at it.

    I think this is partially true, but also remember that beliefs like zealotry or aposolithic tradition have other places to dump faith. That means that buildings are even worse for those beliefs than normal.

    Speaking of: Jesus does the AI think Jesus wanted Orders. Those get grabbed way too often in my games. Is the AI overvaluing their uniqueness?

    ----

    The values I think are correct:

    Gold: 1 per 1, 10 cap
    Food: 1 per 1, 15 cap
    Science: 1 per 2, 20 cap
    Faith: 1 per 1, 8 cap
    Culture: 1 per 2, 8 cap
    Production: 1 per 2, 10 cap

    Reasoning:

    Gold vs Production: Gold comes online faster, but 10 production is better than 10 gold.
    Food: It was weak at 10 cap. 15 will make it more commonly picked compared to Mandirs.
    Science: I think a very high cap here makes sense. This will make it a better long term investment than culture, and allows it to compete with Mastery and Synagogues as a more generalist solution. Both can outperform it.
    Faith: Fast faith is unlike anything else. A good yield at the time that turns useless quickly. Has a spot in some interesting strategies. Never should be picked without a plan, and thus is capped low to avoid a huge window for abuse.
    Culture: I think a lower yield cap is the best balancing spot. It allows Mosques to outperform them in the long run.

    All in all I think that Faith and Culture are the hardest to balance. Once culture is set science become hard to balance, because it's worse than culture or production, but better than gold or food.

    I'm highly open to feedback on these.

    The best way to put it is that I think there's a ~19% chance I'm spot on, a ~38% chance that people will decide on another methodology (changing something I think should be 1 per 1 to 1 per 2 for example) a ~38% chance I'm off by give or take 25% on some of the numbers and a 5% chance I'm way off.

    I mostly say that to give you guys some insight into how my brain works. I always take the chance that I could be wrong into account, and when I give an answer it's not always with the same degree of certainty. Text can make that hard to communicate, so I figured I'd be a dork and type it out.
     
    Enrico Swagolo likes this.
  6. Gazebo

    Gazebo Lord of the Community Patch Supporter

    Joined:
    Sep 26, 2010
    Messages:
    17,848
    Gender:
    Male
    Location:
    Little Rock
    That’s fine - but a simple ‘gold should be x, capping at y’ is more effective than two paragraphs about why I’m wrong to touch this because the AI is made of beans. Just saying.

    You’ll be able to achieve the same effect with Percent as with Halved. We’re at a point where every memory element matters and two tables that achieve the same thing don’t need to exist.

    G
     
  7. Gazebo

    Gazebo Lord of the Community Patch Supporter

    Joined:
    Sep 26, 2010
    Messages:
    17,848
    Gender:
    Male
    Location:
    Little Rock
    ‘Play with the Religious AI code’ is not a solution. We’re in diminishing returns territory here which is why I’m bringing up balance. If I was reasonably sure I could improve the AI beyond its current capacity without a hundred more man hours of debugging I would not have brought this up. So work with me here.

    G
     
  8. YukiN

    YukiN Prince

    Joined:
    Sep 3, 2018
    Messages:
    361
    Location:
    Singapore
    Those are some very high caps for the 1/2s, not in terms of yield but in terms of population requirement. Even 16 population is fairly high for renaissance, let alone 30 or 40. The caps are unlikely to differentiate the 1/2 yields apart enough.
     
  9. Stalker0

    Stalker0 Baller Magnus

    Joined:
    Dec 31, 2005
    Messages:
    6,253
    These look like a reasonable starting ground. I could see faith and culture capped at 7, but that's gut feeling, neither I feel so strongly about. I will say these numbers have the look of balance between them. All of them have a reasonable niche in this model.

    In terms of Science, I do agree 20 is a high number that almost never will get to. Maybe 2 for 3 with a lower cap would be reasonable. The 15 for food I'm fine with. If you are going the food belief you are planning for large cities anyway.
     
  10. Enrico Swagolo

    Enrico Swagolo Deity

    Joined:
    Jun 10, 2013
    Messages:
    2,379
    I also like ElliotS solution.

    I fail to see a problem. So the AI cannot properly work with religious buildings and does better with follower beliefs, but Humans find them both balanced and any drastic adjustments to nerf yield per follower beliefs will make buildings vastly superior. You should obviously side with the human people guys here unless you're secretly a robot, especially since having the choice be balanced for Humans with human settings means that new VP players won't potentially take stuff that is bad for them because it's designed to be balanced for AI vs AI, not for human hands. Like that 1 Faith for 4 pop... by the time you're getting anything out of this, Faith costs are 3x/4x of what they were by the time you took the belief. Someone could get pummeled hard by the AI if he took something bad like that.

    EDIT I think I've misread
     
  11. Gazebo

    Gazebo Lord of the Community Patch Supporter

    Joined:
    Sep 26, 2010
    Messages:
    17,848
    Gender:
    Male
    Location:
    Little Rock
    I think @ElliotS 's are good but culture and faith are still a bit too high - Culture especially breaks games for the AI (I've seen AI with that follower which have 100x more culture than anyone else before).

    Wut. I'm not asking the world - I'm just asking for everyone to help me retune this a bit to be just a little less potent.

    G
     
  12. Deadstarre

    Deadstarre Expert

    Joined:
    Nov 1, 2015
    Messages:
    960
    Location:
    New York
    not sure im completely following you. theres like 4 tables right now between yours and vanilla that basically do 2 different things (raw yields and % per followers) so like if you could make that more efficient while still keeping the ability to scale by pop (1 per 1 or 1 per 2) in a single table thatd be great, i just wanted to check if there would still be a table for both % and for raw yields. differentiating between % and raw yield tables are already all over the place, always useful to have both options.
     
  13. Stalker0

    Stalker0 Baller Magnus

    Joined:
    Dec 31, 2005
    Messages:
    6,253
    I wouldn't be opposed to a 7 or even 6 cap for those if we think we need it. They are still a big shot in the arm early on, they would just taper off quicker.
     
    Enrico Swagolo likes this.
  14. Gazebo

    Gazebo Lord of the Community Patch Supporter

    Joined:
    Sep 26, 2010
    Messages:
    17,848
    Gender:
    Male
    Location:
    Little Rock
    Sorry I misunderstood you - yes, both flat and mod 'halved' tables are becoming percents. The way it works is that you actually need both the 'percent' table and the 'perfollower' table, as the 'percent' table uses the 'per follower' value as the max for the belief. i.e. To get 1 gold per 2, capping 20:

    Code:
    INSERT INTO Belief_MaxYieldPerFollowerPercent
        (BeliefType, YieldType, Max)
    VALUES
        ('BELIEF_FEED_WORLD', 'YIELD_GOLD', 50);
    
    INSERT INTO Belief_MaxYieldPerFollower
        (BeliefType, YieldType, Max)
    VALUES
        ('BELIEF_FEED_WORLD', 'YIELD_GOLD', 20);
     
  15. Gazebo

    Gazebo Lord of the Community Patch Supporter

    Joined:
    Sep 26, 2010
    Messages:
    17,848
    Gender:
    Male
    Location:
    Little Rock
    @Deadstarre's post reminds me that we do have % modifier for followers as well - we could 'hybrid' some of the beliefs so that they're a blend of % per follower and + per follower.

    G
     
    Gokudo01 likes this.
  16. Deadstarre

    Deadstarre Expert

    Joined:
    Nov 1, 2015
    Messages:
    960
    Location:
    New York
    it is kind of confusing to talk about. i do understand what you did right there, but im not sure if you just demonstrated what will be 1% gold per 2 followers (cap 20%) or 1 flat gold per 2 followers (cap 20 gold). your 2nd post makes me think it was the flat gold, which does maintain current functionality but in a more efficient and flexible way, which would be great. so i think... sounds good. :thumbsup:
     
  17. Gazebo

    Gazebo Lord of the Community Patch Supporter

    Joined:
    Sep 26, 2010
    Messages:
    17,848
    Gender:
    Male
    Location:
    Little Rock
    Haha the other table is called Belief_MaxYieldModifierPerFollowerPercent. :crazyeye:

    G
     
  18. ElliotS

    ElliotS Warmonger

    Joined:
    Jun 13, 2013
    Messages:
    2,704
    Gender:
    Male
    Location:
    Chicago
    The 20 is meant to be too high for renaissance and even some by game end. It makes the belief better for tall. All of them had intended dropoff points where they stop increasing, and not all of them are meant to be at max in all or any cities when you get them or by Renaissance.

    Then lowering the cap is the right play I guess, but that stat is absurd lol.

    Is there some way to make buildings more AI friendly? That seems to be the issue. Maybe have them cost 0 faith but produce negative faith? (Kinda like faith maintenance.)

    That would also nerf cases where you get many extra buildings.
     
  19. chicorbeef

    chicorbeef Emperor

    Joined:
    Dec 26, 2017
    Messages:
    1,359
    Gender:
    Male
    I like all of them, with the exception of Culture, I think Culture could be 1 per 3, cap at 10. Culture does not suffer from inflation so much in the way other yields do, getting policies faster speeds up your whole game.
     
    Owlbebach and phantomaxl1207 like this.
  20. CppMaster

    CppMaster King

    Joined:
    Feb 13, 2018
    Messages:
    927
    Location:
    Poland
    Per-yield followers will be even better for AI, because they have growth bonus. According to Owlbebach's scrutiny, AI growth bonus doesn't make much difference in terms of overall strength, but may break balance in cases like this.
    @Gazebo have you considered removing AI growth bonus with slightly improving other bonuses?

    https://forums.civfanatics.com/threads/removing-growth-from-ai-handicap-list.635134
     
    Owlbebach likes this.

Share This Page