Good call I didnt even consider that feature because how insignificant it is as human player.
3.6 AI-only Game Statistics
- Thread starter L. Vern
- Start date
pineappledan
Deity
Sivs that we are 95% sure are strong in AI hands:
- Ottomans - This doesn't surprise me in the least. Users report he doesn't feel overly strong, but my impression with him is that he just needs to make it to Renaissance and then he is unstoppable.
- proposal: reduce Ottoman's instant yields from 150 to 100, scaling with era.
- Japan - Japan has consistently been near the top for their ability to convert warring into CV. The recent conquest improvements have affected them predictably. They have also received numerous buffs over the last few versions, making their bushido promotions stronger and giving them to more units. Their early tech path has been repeatedly buffed (numerous barracks buffs, bronze working buffs, etc.). Not only was this civ doing fine before, it feels as though some of the recent changes to the early game were tailor made to enable and accommodate Japan.
- proposal: revert the change that gave Bushido to boats. It was an unnecessary steroid to a civ that already was doing well.
- Better yet: rework the UA entirely. Maybe give Japan's GWAM bonus to France, so that only 1 of these civs is giving this type of bonus. The +1
on some buildings in the UA is possibly the most creatively bankrupt and uninteresting bonuses given by any civ's kit
- Austria - very powerful for both AI and humans
- proposal: remove WC vote from marriages
- Arabia - Consistently one of the strongest AI civs, and users self-report he is one of the easiest civs to win with. There is almost nowhere to go with this civ, because his components are all at the bottom of their possible power. His UB is comparatively weak, and his stacking bonus is already +1. There is no further down to go from 1 except 0. The problem with Arabia is that HEs are too common, too frequent, and too early, and that his UA greatly increases the frequency of HEs by increasing GP births.
- Proposal: remove the %
GP completion bonus from HEs and increase the yields per HE or strengthen the UB. Arabia's UA makes HEs both more frequent, and more rewarding; It should only do 1
- Proposal: remove the %
- Greece - not surprised. CS decay was greatly increased overall, and Greece suffers exactly 1/2 of these increased penalties to CS influence as everyone else at all levels.
- Proposal: reduce the influence decay reduction to -33%
- India - not surprised in the least. The power of the UA was almost tripled in the latest version. India was weak before, but he wasn't so weak he needed that level of help
- proposal: reduce the scaling base pressure from +1 per follower to +1 per 2 followers
- Better yet: change the UA entirely so that it doesn't boost a completely passive game component, and give India a more active spreader ability, like boosting the spread actions of GProphets
- Germany - not surprised. The UA is ok, the UU is good, the UB is possibly the strongest in the game
- proposal: reduce the global %
modifier on the Hanse from +5% per city state TR to +3% per City State TR
- proposal: reduce the global %
- Poland - One of the most consistent civs in these AI tests accross multiple versions. He has never broken into top 3, but he has always been at least top 10.
- proposal: remove the 100 on construction of the Ducal Stable
- proposal: remove the 100
- France - We have been talking about how bad France is for over a year. This civ's UA needs to be reworked. It's not even particularly fun.
- proposal: Either give France an entirely new ability, or give them Japan's version of the GWAM trigger. Either way, the city conquest bonus needs to go.
- Polynesia - a bit surprised to see Polynesia here, even after we added more on tiles to
converters. this civ has never felt lacking to me. they also were recently buffed- proposal: no change for now. No ideas.
- Venice - Venice is gonna venice; he was top 5 back in v3.2, and now that trade route gold has been reduced, he's down at the bottom. They have changed so much so many times it makes my head spin to suggest changing them again. But, I think one of the biggest problems with Venice right now is that it only has 1 city that can produce GPPs. This either needs to be changed for Venice, or they need some way to compensate for it, because right now they aren't competitive as a CV civ, even though they have an entire wonder dedicated to going that route.
- proposal: Allow Venetian puppet cities to generate GPPs
- proposal: Allow Venetian puppet cities to generate
- Spain - In addition to being weak, player feedback has also described them as over-complicated and hard to use. The most obvious shortcoming of their kit, however, is that the faith costs for military units are too high to make their special ability to purchase ships worth using.
- proposal: greatly reduce the faith costs of naval units.
- Carthage - Carthage has moved from the top to the bottom as trade route gold calculations have changed. If Trade route gold is at a place where people are happy with it, then it might be time to change Carthage's components
- Proposal: needs to be discussed further. Carthage is also targeted by a passed, but unimplemented proposal to change triremes
- Huns - in addition to being weak, users have also reported that playing the Huns is generally not very fun. Their War Weariness bonus is hard to keep track of, and is too subtle right now. Their Eki takes too long to build and is a pain to set up. Their unit capture mechanic gives you lots of worthless, heavily damaged units that feel bad to lose as soon as you converted, and feel even worse to move back to your territory to sell or heal up for 10 turns. In general, the Huns demand a lot of work from the player with relatively little return.
- Proposal: Double the War Weariness inflicted bonus and reduce the build time on their Eki
- Better yet: change how the Huns' unit capture ability works, so that it gives less frequent, but better units. The Huns could automatically convert partisans when razing a city instead of capturing killed units.
- Denmark - I'm not convinced the AI knows how to play Denmark, and I worry that any buff given to them would just make them OP in player hands
- proposal: No change
- Russia - I'm genuinely shocked to see Russia so close to the bottom, but I suppose their UA and UB have both been nerfed pretty hard lately.
- Look at buffing the Ostrog. maybe just increasing its base yields, or restoring their bonus to border growth.
- Mongolia - I don't think that the flanking bonus for skirmishers is particularly strong, and Mongolia's tribute bonus is suffering from the underlying mechanic being weak
- Proposal: change the tribute bonus from +100% of tribute yields to giving an additional 15% of yields from tribute as All Yields (
), and buff the tribute mechanic
- Proposal: change the tribute bonus from +100% of tribute yields to giving an additional 15% of yields from tribute as All Yields (
- Songhai - this civ's UA should be radically changed anyways; they're just the store-brand version of the Iroquois' abilities.
- Proposal: change Songhai's UA so it does something else. Discuss.
- Persia - Golden Ages are too easy for civs in general. I think the best way to buff Persia is to nerf Golden Age generation for everyone else.
- Proposal: greatly reduce the
GAPs generated by GArtists, and raise the cost of all Golden Ages.
- Proposal: greatly reduce the
Last edited:
I think he fall off outside of pangea, on pangea or maps with even less waters he is strong enough to not have in the game.
The AI hasn't learned to attack from water using embarked units while under a ship that's not attacking to conserve health.
Legen
Emperor
- Joined
- Sep 13, 2015
- Messages
- 1,456
Strongly disagree with the approach. I find that what has been pushing AI Japan's winrate is the abnormally high number of scientific victories they get, despite the civ's kit not being designed for that. In this latest simulation, Japan got proportionally more scientific wins than Korea, Babylon, Maya and Assyria, who are tailored around such victory, and one of Japan's scientific wins had them matched against Korea and Babylon. Other recent simulations also had Japan getting a good number of scientific victories as well, comparable to dedicated scientific civs, so I don't think this one is an exception.Japan - Japan has consistently been near the top for their ability to convert warring into CV. The recent conquest improvements have affected them predictably. They have also received numerous buffs over the last few versions, making their bushido promotions stronger and giving them to more units. Their early tech path has been repeatedly buffed (numerous barracks buffs, bronze working buffs, etc.). Not only was this civ doing fine before, it feels as though some of the recent changes to the early game were tailor made to enable and accommodate Japan.
- proposal: revert the change that gave Bushido to boats. It was an unnecessary steroid to a civ that already was doing well.
- Better yet: rework the UA entirely. Maybe give Japan's GWAM bonus to France, so that only 1 of these civs is giving this type of bonus. The +1
My attention is on the Dojo. Aside from giving +3
science over the Armory, the Emperor AI also has a special benefit from the leveling mechanic, as a city with Barracks and Dojo would get a unit to level 4 immediately building it for this AI, instead of level 3 that humans get. This means that, instead of gaining 5 after creating a military unit, AI Japan gets 21 . Add to it that AI units also earns more experience from combat (iirc +60% on Emperor), and the AI gets a lot more science than a human could from this mechanic.We already saw how Japan's leveling mechanic could get out of control with the AI handicap bonuses in the past, when Gazebo experimented replacing Japan's GWAM UA with other mechanics. One of these experiments had the leveling mechanic moved from the Dojo to the UA (and, therefore, available from turn 1). Deity AI Japan easily dominated, with someone reporting AI Japan getting to Medieval when other AIs had just got to Classical. And human players could sometimes get a taste of that with a scout and lucky ancient ruins. Since Japan has had an easier time with Domination lately (which they're intended to excel at), it is likely that this mechanic is turning warring into science better than even Assyria's kit.
I've been thinking of replacing the science on leveling for something else, so that Japan stops performing like a dedicated scientific civ in AI hands. One idea I'm considering is to remove the +3
on the Dojo and replace the science on leveling for faith instead; this way, the Dojo's leveling mechanic mirrors the yields Japan gets from military and defensive buildings, establishing a more coherent theme. Older versions had the Dojo giving on leveling, so faith is not a radical design shift for the Dojo.I don't think further copying Japan's UA would help France, and could make it worse. Japan had many patches in which the GWAM didn't translate into good AI performance, and there were patches in which some players would argue France's GWAM version was outright better than Japan's. France also has no directed support for Great General/Admiral generation, notably lacking Japan's focus on an army composition that prioritizes units that generate plenty of Great General/Admiral points.
I find the main issue with France comes from its UA design piling support for an effect that no longer exists, which resulted in an all-or-nothing UA. Unmodded BNW France had double theming bonuses as its UA, and VP first added support to it with an ability to steal great works on city conquest. This first support wasn't without flaws, as the opponent may not have yet any great work to steal; this was then adjusted to gain some cultural benefits in such case, just like Assyria. Later on, the great work theft was replaced with GWAM on city conquest (when Japan was undergoing experimentation with a Sakoku UA), and the culture on conquest was no longer tied to failing to steal a great work (moving it away from its original intention). At some point, the doubled theming bonus was removed from the UA, meaning France's design direction went from supporting double theming, to being practically all about city conquest.
I think there's a good reason why double theming bonuses was a recurring request back in the France balance thread, and bringing it back can realign France's design into a coherent direction. I proposed before a double theming UA that tried to do that, but it wasn't sponsored. I'm thinking of proposing something like it again.
Last edited:
IMO using this data to "balance" the game without decoupling it from tendencies/personalities (for units, policies, war, whatever) is a silly idea.
Even then I would be cautious about balancing based on AI games. The AI gets numerical bonuses that interact with other bonuses in ways that a human won't experience (although I couldn't find what difficult these games run at). It's certainly desirable for the AI to play equally well with all civs, but it's not going to be achievable without compromising the fun of actual players.
Even then I would be cautious about balancing based on AI games. The AI gets numerical bonuses that interact with other bonuses in ways that a human won't experience (although I couldn't find what difficult these games run at). It's certainly desirable for the AI to play equally well with all civs, but it's not going to be achievable without compromising the fun of actual players.
pineappledan
Deity
It generates discussion about what civs also feel too strong or weak in human hands. It also provides much better quantitative data than some human self-reporting. We don’t have to balance civs perfectly in accordance to these games, but it helps guide discussion.
I don’t necessarily agree that it’s “silly” anyways. On a standard game, 7/8 of the players in a game are AI. We need to make sure that the AI can handle their civs competently against other AI, because that is who they are playing against the majority of the time. If the outcomes of those AI matchups are too predictable then it will be a bad human experience too.
These games are played on emperor.
Perfect balance isn’t achievable or desirable, but pruning the extreme top performers and bottom performers seems reasonable, especially when player feedback aligns with the AI-only results.
pineappledan
Deity
Or instead of making Japan more of aI've been thinking of replacing the science on leveling for something else, so that Japan stops performing like a dedicated scientific civ in AI hands. One idea I'm considering is to remove the +3
Civ, we could embrace that he has sources of in his kit that he can use to achieve that victory. We have relatively few civs overall that give decent bonuses; why would we replace one for a more common bonus?If we have to make a decision between removing Japan’s science bonuses or his religious bonuses, then the clear loser is his faith bonuses. Those
on building class bonuses should die in a fire; they would be lame even as a policy bonus, but are catastrophically sad as a UA bonus.Then you remember things differently than me.
There was tacit acceptance for that proposal, but not enthusiasm. People agreed that France’s current UA is terrible, but the reception we got for proposing to bring back the theming bonus was not great. Especially in the context of the work involved by someone else to code it. It’s a lot of work to implement for a UA ability that no one particularly wants.
Even if it were coded, I’m not sure it would stay. People have pointed out pretty fair criticism about how late and slow the theming bonuses are to generate, and how generally irritating the mechanic can be. The first theming bonus isn’t until late classical, for instance.
The only direct support Japan has is the Great Generals II promotion on his samurai. France also has a melee UU. You could simply give France direct support for the ability in a few ways, any of which would be easier to implement than the theming bonus stuff:
Esprit de Corps
- +10% Damage and +1 GGeneral/GAdmiral Point for each subsequent attack against a single target during a turn.
- When a Great General/Admiral is born, receive Great Artist, Writer, and Musician Points in your Capital
It would be interesting to do the same analysis with Random Personalities enabled.
It could be dominated and I image the balancing would include it, since the default way of playing is with those personalities.
Why not on Chieftain (or Warlord?), so it's bonus neutral?
Why is it not desirable? It's better to have the game more balanced than not.
N.Core
1st Class of Neoa
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?
pineappledan
Deity
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
Last edited:
It comes down to how we want the civ to play. Do we want Japan to be a science heavy civ?Or instead of making Japan more of a
It can be argued that science is a double power boost for a warmonger. Traditional warmongers normal lack in tech, and so have to use their superior combat boosts often against superior units. Japan gets the best of both worlds, a strong bonus and high quality units, which is likely helping their warmongering quite a bit.
I respect your desire to completely change Japan, and I'm sure you will want to put in a proposal for some of those ideas. But for those looking for a simplier balance change that doesn't completely change japan, I do think a simple science nerf (no replacement just pure nerf) is the simplest way to approach it.
pineappledan
Deity
I think France / Japan balance and design should move to a separate thread.
There is a bigger discussion to be had about what we want both of those civs to do, because right now they are doubled-up on a lot of similar mechanics, rewards, emphases, and components, to the detriment of both.
There is a bigger discussion to be had about what we want both of those civs to do, because right now they are doubled-up on a lot of similar mechanics, rewards, emphases, and components, to the detriment of both.
N.Core
1st Class of Neoa
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.I would be quite interesting to know how often AI leaders have a religion (and when they acquired it, either through foudning or conquest).
We have often questioned the strength of religion. If religion is "super strong" than it may be the civs that are able to found (or conquer a holy city) have an innate leg up versus other civs, regardless of their bonuses on the surface.
That said, only India has a major religious bonus, with Japan a minor one. So it doesn't seem that strong religious play is a key component of winning.
We have often questioned the strength of religion. If religion is "super strong" than it may be the civs that are able to found (or conquer a holy city) have an innate leg up versus other civs, regardless of their bonuses on the surface.
That said, only India has a major religious bonus, with Japan a minor one. So it doesn't seem that strong religious play is a key component of winning.
pineappledan
Deity
Arabia has a minor religion boost too (2 on bazaar)
- there are 3/7 civs in the “too strong” bracket that have some sort of Boost (Japan, India, Arabia)
- 2/11 of the “too weak” bracket have any bonus (Spain & Mongolia).
- There are 7 other ‘religious’ civs with the 25 civs in the middle-tier (Ethiopia, Morocco, Byzantium, celts, Siam, Aztec & Mayan). I count Siam because if he meets a religious CS at any point he gets a free 2 for 10 turns. I don’t count Egypt, because the 1on his burial tomb is way too late to be any help with founding; it’s just flavor.
In other words, 43% of the overpowered civs have a religious component, 28% of the balanced civs have a religious bonus, and only 18% of the underpowered civs have a religious component.
My working theory is that founding has a very strong impact on winning. I will assume until proven otherwise that the games that are won by any civ are the games in which they founded a religion.
on bazaar)- there are 3/7 civs in the “too strong” bracket that have some sort of
Boost (Japan, India, Arabia)- 2/11 of the “too weak” bracket have any
bonus (Spain & Mongolia).- There are 7 other ‘religious’ civs with the 25 civs in the middle-tier (Ethiopia, Morocco, Byzantium, celts, Siam, Aztec & Mayan). I count Siam because if he meets a religious CS at any point he gets a free 2
for 10 turns. I don’t count Egypt, because the 1on his burial tomb is way too late to be any help with founding; it’s just flavor.In other words, 43% of the overpowered civs have a religious component, 28% of the balanced civs have a religious bonus, and only 18% of the underpowered civs have a religious component.
My working theory is that founding has a very strong impact on winning. I will assume until proven otherwise that the games that are won by any civ are the games in which they founded a religion.
Last edited:
