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

Thoughts on A/B/C bonuses

Discussion in 'General Balance' started by Milae, May 22, 2020.

?

What do you think of A/B/C bonuses?

  1. Like them.

    8 vote(s)
    22.9%
  2. No strong opinion.

    4 vote(s)
    11.4%
  3. Don't like them.

    12 vote(s)
    34.3%
  4. Don't like the production part.

    16 vote(s)
    45.7%
Multiple votes are allowed.
  1. ElliotS

    ElliotS Warmonger

    Joined:
    Jun 13, 2013
    Messages:
    2,861
    Gender:
    Male
    Location:
    Chicago
    Part of the thing is that I think a lot of players differ on what they want in difficulty. If the optimal difficulty is for AI performance to try to match player performance from the start to the end, ABC will do it best because players DO snowball.

    If you want AIs to progress at a set rate in a relatively close clump with a few dropping off as they get beaten up, we can probably make that happen better with a scaling % bonus instead of ABC.

    That makes their performance less responsive to their own actions and the gamestate, and makes them more of a bar saying "You must be this tall to win" instead of feeling like real opponents however.

    Real players snowball. Real players that are doing good get and maintain a strong lead. Real players that are getting kicked into the dirt are not great at coming back and perform much worse than a player flourishing in a wonderland of great people and wonders.

    I prefer AIs that mimic that behavior a bit better, within reason. So while I can support some efforts to keep the top-end in line with what a player really going off can do, I don't want to switch to a system that makes the AI so much less dynamic and responsive to the gamestate.
     
    Ziad, vyyt and 2506 like this.
  2. Amask

    Amask Oogala Boogala

    Joined:
    Feb 7, 2002
    Messages:
    776
    Location:
    Toronto
    I have a game currently where the runaway (who's 6 techs ahead of 2nd place, and 10+ ahead of average level) keeps getting DOW-ed by every single neighbor over and over. Throughout the course of the game he only lost one city out of eight. With the exception of that one war I would imagine the runaway "won" every other war (better units and being on the defensive means he probably did more damage than he received), and got showered with more bonuses. I didn't have logging enabled, so I can't confirm this for sure, but it seems logical.

    He isn't accomplishing anything with those defensive wars, yet he keeps getting boosted further and further ahead.
    This is in contrast to your claim that a human can get more out of a war than an AI and therefore AIs should get boosted for winning wars.
    Offensive wars, sure. Defensive wars, not at all.
    He keeps getting dogpiled (he was in a war with all five neighbors simultaneously at more than one point), and it only benefits him.
    That's completely nuts.
     
  3. Stalker0

    Stalker0 Baller Magnus

    Joined:
    Dec 31, 2005
    Messages:
    6,804
    Actually I just checked the Wiki, and it looks like this is already the case. All the values it notes there use the "average era"
     
    Ziad likes this.
  4. Stalker0

    Stalker0 Baller Magnus

    Joined:
    Dec 31, 2005
    Messages:
    6,804
    The answer to this....is maybe. AIs get a bonus for a winning a war, ONLY if there warscore was 25 or more. So if it was just two sides potshotting each other, no bonus. If the runaway was dominating the wars, then yes he is getting bonuses. You can hover over the leader portrait to see generally what kind of warscores the main guy is getting.
     
    vyyt likes this.
  5. Ziad

    Ziad Emperor

    Joined:
    Nov 11, 2013
    Messages:
    1,880
    Location:
    Lebanon
    Oh I get that...the numbers were just an example. Stalker0 phrased it better.
     
  6. Stalker0

    Stalker0 Baller Magnus

    Joined:
    Dec 31, 2005
    Messages:
    6,804
    Here is the main list of what things the AI gets bonuses on, so I'm going to give me thoughts on each:

    • Enters a new era (3x normal bonus; Food, Production, Gold, Golden Age Points, Science, Culture)
    -- This is the biggest boost for the AI, and it could be argued one that tends to push the runaways ahead. That said, it is also one of the most consistent bonuses (it happens every game 7 times). What we could do here is simply give all AIs their era boost when the average era increases. The AIs are still scaling and getting their big push, but more as a group, so no AI doesn't overshoot another.
    • Founds its original capital (Food, Production, Gold, Golden Age Points)
    -- The diety players have brought up a lot of concern about early religion being so difficult because the AI defacto builds a shrine on Turn 1. With that in mind, I think its reasonable to remove the Production and Gold bonuses here, but leave the other bonuses intact. The growth at this point in the game is still a good bonus.

    Another option here is to remove the capital found but replace it with a bonus upon finding your Pantheon. So overall bonus is the same, but just delayed a bit to give players a chance to settle themselves.
    • Founds a new city, other than its capital (Food, Production, Gold, Golden Age Points, Science, Culture)
    -- On one hand, it seems "unfair" that wider AIs get more bonuses. That said, most of the runaway complaints on the forums right now are isolated Tradition civs. Further, city placement is a hard thing for AIs to optimize, and knowing how to build more cities to compensate for the penalties is something the human is much better at. So I think its worth keeping this as is.
    • Wins a war (warscore 25+) (Food, Production, Gold, Golden Age Points, Science, Culture)
    -- Warlike AIs do not seem to be the runaway problem at the moment, so I'm inclined to leave this be. We could tweak the warscore number a bit if the "defensive war" problem pops up.
    • Starts a Golden Age (Food, Production, Gold, Science, Culture)
    --This one is interesting to me. My general guess is that the only time this really magnifies runaways is if an AI falls into unhappiness early. Otherwise, all of the other GAP bonuses means its less the difference of how many GAs you get, and more of when they happen. However, a very unhappy AI could lose all of his GAP, and therefore fall behind. I could take it or leave it, my guess is its not a major contributor until the late game when everyone is permanently GAing.
    • Constructs a World Wonder (Gold, Golden Age Points)
    --The original major complaint, and has gotten heavily nerfed (only provides gold bonuses now). Still strong but much more reasonable. Probably could knock the GAP bonuses if we wanted to take it down one more notch (wonder hoarding Tradition civs already get a lot of GAP, no need to throw more on the pile).
    • Generates a Great Person (Gold, Golden Age Points)
    --Similar to Wonders, this one has already been pretty heavily nerfed. That said, its likely a large portion of a Tradition AIs gold reserve, as GPs are quite common. I still wonder if any bonus here is necessary, as honestly this is the most "mundane" of the activities listed here.
    • Completes an antiquity site dig by building a Landmark or extracting an artifact (Gold)
    --Considering that there is both a pretty wide difference in the number of landmarks within a person's territory and that artistry can literally double this number...I would be inclined to drop it. Getting works and landmarks is already a good bonus.
    • Completes a trade route to another civilization or City-State (Gold)
    --This one to me is highly variable and random. I also don't know if the AI knows how to optimize TRs well (aka is one AI getting 20 turn TRs whereas another gets 50+ turn ones just out of pure luck). Further, this doesn't seem to scale well with map size. I say kill this one.
     
    AndreyK, Rhys DeAnno and vyyt like this.
  7. InkAxis

    InkAxis Warlord

    Joined:
    Feb 24, 2020
    Messages:
    276
    Gender:
    Male
    I'd just say that they should be tweaked for starting your capital or at least lowered for shrines. Getting a monument a couple turns earlier means more culture which is really good but doesn't affect other players. Meanwhile getting a shrine really early means getting a religion earlier, and even a couple turns earlier can mean the difference between getting a religion which changes the entire game. Whereas culture bonuses you can at least catch up. So I would say at the least change it so it doesn't affect shrines.
     
    Vhozite and vyyt like this.
  8. Amask

    Amask Oogala Boogala

    Joined:
    Feb 7, 2002
    Messages:
    776
    Location:
    Toronto
    Good point. I know the main guy did have at least 20 warscore at certain moments (not sure about 25), but I wasn't really paying close attention throughout the entire game. I definitely saw him damage neighbors' cities many times, so he wasn't just camped up in his land.
    At the very least he's swimming in great generals. GP birth boost is not as substantial of a boost as winning wars, but a boost nonetheless.

    The AI in question in this game is Pacal, and this time I turned random personalities on. It's actually kind of fun facing a psychotic Pacal. He's the one who started the first few wars, then everyone turned on him, but he doesn't care. He just dropped the first nuke of the game on a neighbor.

    Anyway, my problems with the ABC system are that's it's more random than I think it was designed to be, and some aspects are unnatural.
    The era transition could be the biggest culprit, as you pointed out. It's a boost that comes earlier to some than to others, more or less randomly. And then that scales up their other boosts. And obviously the earlier this happens the better.
     
    Rhys DeAnno likes this.
  9. azum4roll

    azum4roll Emperor

    Joined:
    Jul 17, 2018
    Messages:
    1,072
    Gender:
    Male
    Other than food and production, I think all of the other yields should be a fixed amount independent of the number of cities, and the culture should not give city expansion points. At the same time, the settling AI should be modified to ignore suboptimal spots when going tall, after securing their monopoly and the surrounding land of the capital. That way we can see actual tall AIs, and tall/wide performance would not be skewed in AI games (now AIs with low amount of cities always fall behind, Tradition or not).

    The cumulative ABC bonus for settling is currently scaled quadratically (linearly with both number of settles and number of cities), so the initial settling spree kickstarts the snowballing of AI civs. With the above change, the total ABC bonus scales linearly with number of settles, so the early game progression should feel smoother.

    The production yield should not be usable on World Wonders, World Congress Projects and Shrines. It can be saved to the excess production pool if the city is currently building one of those.
     
    InkAxis likes this.
  10. tothePAIN

    tothePAIN King

    Joined:
    Dec 31, 2018
    Messages:
    617
    Gender:
    Male
    These are good suggestions. It's frustrating to see an AI finish 3 world wonders in a turn or have 15,000 production for a world project.
     
    Rhys DeAnno likes this.
  11. Recursive

    Recursive Emperor

    Joined:
    Dec 19, 2017
    Messages:
    1,771
    Gender:
    Male
    Settler AI was modified in this version (and modified some more for next version). RE: Yields besides food/production - this is already the case.

    Saving it to the excess production pool is not friendly with the existing code - and I think the overflow amount is limited.
     
  12. InkAxis

    InkAxis Warlord

    Joined:
    Feb 24, 2020
    Messages:
    276
    Gender:
    Male
    There are a lot of other ways this could be implemented. For instance the AI would be forced to switch to any production other than shrines/wonders/WC projects. Or it could just directly give production to certain buildings, basically the same effect.

    Also, another thing I've been been thinking about is perhaps the bonuses should be slightly negatively scaled per event. This would round things out better as people have mentioned that it can be more random than expected. It would be very slight, like -10% after several events in a row, and it would reset/decay after a while. This would help for sudden random spikes the AI has. If there were more of these (for instance one for getting a social policy) but each have less effects would help. So that way it would also be more smooth.
     
    Last edited: Jul 22, 2020
  13. Gidoza

    Gidoza Emperor

    Joined:
    Jul 26, 2013
    Messages:
    1,270
    You know I'm not quite sure why this was never apparent to me before, but I just now realized how these bonuses affect newly-found cities for an AI. The situation? An AI founds a city near me. That's fine. With a soon declaration of war, I kind of expect to just knock it off quickly due to low defence even with a defender. BUT, because of the super production bonus - the AI instantly builds walls, which nuke one of my units only for the spearman inside to follow up and kill it instantly.

    Bonuses for the AI? Ok sure. But actually changing the landscape of how a war or territory functions??? That's straight-up inappropriate. Maybe if the production allocations to buildings took the place of Gold Investments...the AI should still take the same amount of turns to build its Walls as it would take me to do so after investing. Either that, or make the Production Bonuses into something that's gradual over time - e.g., instead of 200 instantly, make it 10 per turn for 20 turns, or something.
     
    a3kov, Vhozite, vyyt and 3 others like this.
  14. Rhys DeAnno

    Rhys DeAnno Prince

    Joined:
    Feb 22, 2011
    Messages:
    360
    Gender:
    Male
    Yeah the harsh truth is unlike a naked human city which will blow over in a stiff breeze a turn after founding, a new AI city can become quite beefy in a matter of a couple turns. The better counterplay is to declare before it plops, either to snatch the settler, or if it's well escorted, contest its route to the spot.

    Of course, it won't ALWAYS become beefy, especially if it was given some peacetime after settling to build peace buildings with its bonus hammers instead. If it's settled during war or war right after settle though, expect beef quickly.
     
  15. tothePAIN

    tothePAIN King

    Joined:
    Dec 31, 2018
    Messages:
    617
    Gender:
    Male
    This is an interesting suggestion. What is the AI A/B/C bonuses that were given for certain events, settling cities, etc. were spaced out over multiple turns? Is that possible? It would solve a lot of the frustrating events that currently occur.
     
    Ziad and Rhys DeAnno like this.
  16. bahamut19

    bahamut19 Chieftain

    Joined:
    Jun 15, 2020
    Messages:
    3
    My first comment...
    I've been playing VP for over a year and I feel the ABC bonuses are the feature which needs changed the most. The rest of VP and modmods for VP are great work and I'm thankful to the community for being so committed to making a nearly perfect 4X experience. I have a few ideas for the bonuses that I'm not sure has been discussed yet.

    Option 1) Make the ABC bonuses available to the AI based solely on human player activity. If a player settles a 2nd city or builds a wonder, the AI gets their bonus. This will help keep the AI bonuses responsive to player ability. If there is a multiplayer game, ABC bonuses are divided by # of players to keep the scale equal.

    Option 2) someone earlier made this suggestion, but spread the bonuses out proportionately to the skills the AI is already doing, a % distribution. I think these skills should be Science, Production, Culture, Gold, Faith, and Golden Age points. GP points should be excluded as the other bonuses indirectly can lead to GP generation.

    Option 3) scale the bonuses by overall AI strength as compared to the #1 civ in the game, by score rank. Say there are 10 Civs, the #1 civ gets 10% of all available ABC bonuses (1 / 10), the #10 ranked Civ gets the 100% of available ABC bonus (10/10).

    I think if all 3 of these are implemented, snowballing would be extremely difficult for any person or AI, it would scale to player ability and play style, and still give a great challenge for the player.
     
    CrazyG likes this.
  17. InkAxis

    InkAxis Warlord

    Joined:
    Feb 24, 2020
    Messages:
    276
    Gender:
    Male
    Another idea I liked is the pseudo- investment, where the AI production bonus is maxed at around 50% per building. We would have to figure out what to do with processes.
     
  18. Stalker0

    Stalker0 Baller Magnus

    Joined:
    Dec 31, 2005
    Messages:
    6,804
    Since A/B/C adjustments can have major impacts to the game, I am fine with going at this at a slower pace. With that in mind I am going to repeat just one of the suggestions I made before.

    • Enters a new era (3x normal bonus; Food, Production, Gold, Golden Age Points, Science, Culture)
    What we could do here is simply give all AIs their era boost when the average era increases. The AIs are still scaling and getting their big push, but more as a group, so no AI overshoots another because of their tech path choice.


    So what this would do is ensure that the first civs entering Classical don't get an immediate boost of all the key yields which could kick off a snowball. Instead, they have to wait a bit until the majority of civs have made it to classical (aka average era goes up). So this way the AIs are all boosting together. The ones ahead are still ahead...but they don't suddenly get a boost ahead that AIs below them aren't getting.

    But on the flip side, we aren't messing with the frequency or intensity of bonuses...not yet at least. Just a simple timing change to see if this helps smooth out the runaway impact.

    I think this is a good starting point... probably not the end point, but a good test to see how powerful timing changes can be.
     
    AndreyK, cerk, a3kov and 1 other person like this.
  19. CrazyG

    CrazyG Deity

    Joined:
    Oct 14, 2016
    Messages:
    5,356
    Location:
    Beijing
    Welcome!

    This is a good proposal.
     
    Drakle likes this.
  20. ElliotS

    ElliotS Warmonger

    Joined:
    Jun 13, 2013
    Messages:
    2,861
    Gender:
    Male
    Location:
    Chicago
    I don't think any of these suggestions would be needed if production was removed from A/B/C bonuses.

    It would also balance the trees much better. The massive consensus is that tradition is best for AI and Authority worst, and it seems it's because AI gets too much production so tradition has no weaknesses and authority has no strength.

    I don't even know if we need to replace the production that's lost. General consensus is that the top snowballers (all tradition civs) are just too good, so a nerf to difficulty would make peaceful play more viable.

    At the very least wait a patch and see if they need to be compensated. My guess is that they don't.
     
    azum4roll, a3kov, Vhozite and 3 others like this.

Share This Page