Game needs rebalancing. Faith is best food and production is best science.

[PSG]

Production conversion to science should be nerfed. An average Makers Quarter will produce 5 times as much science as a science district does when production to science conversion is on. Angkor Wat should be nerfed by 50% .

The optimal path to win fast is Egypt, Maya, Khmer, Mughals and France. Game ends before the final phase if you get decent enough resources.

The only things you should build are production, gold and wonders. Maybe an occasional farm just to optimize growth and maybe a single research center per city so you can build science buildings that boost research %. Do not worry about stability. Just keep building districts.

Wonders to build: Pyramids, Hanging Gardens (if you can get a second wonder), Mausoleum, Angkor Wat, Machu Picchu, Saint Basils. You can guild Notre Dame, but that just takes away more time building Maker's Quarters. This gives you enough food and research. Every district with the Mausoleum is 2 research and this alone will let you finish all important techs every era until you can turn on production = Science when you become France.

Don't worry about stability. As long as you have been a beast producing MQs and some gold production you will be max stability albeit pink and dropping 3%/turn. Beeline for micro-biotics and the other tech that gives you +1 stability per population and then beeline for Fusion. After that go after neural implant. At this stage all your cities will turn 100% red stability. After that it's mop up the remaining tech and game over.

 

[DrCron]

The game does indeed need a lot of balancing (production is definitely OP), but I'm not finding it that easy to pick early-mid wonders on HK difficulty. Especially Angkor Wat, which the AI seems to prioritize in many of my games.

 

[NombreyApellido]

it all stems from the shody AI. it just can't play the game so everything has to be dumbed down to this mess of bonuses and costs bloating for it not to choke on simple tasks

balancing the game to make some sort of sense and expecting this "AI" to play it would be like feeding rybka chess them pac-man ghosts

just try to enjoy it for what it is

 

[PSG]

it all stems from the shody AI. it just can't play the game so everything has to be dumbed down to this mess of bonuses and costs bloating for it not to choke on simple tasks

balancing the game to make some sort of sense and expecting this "AI" to play it would be like feeding rybka chess them pac-man ghosts

just try to enjoy it for what it is

The Ai is really good at founding a city next to one of mine and spawning 8 units immediately.

 

[NombreyApellido]

it's so good it almost feels like it is playing a different game

 

[PSG]

At least the Ai builds actual infrastructure in Humankind as opposed to Civ games.

 

[Don Andre]

Not to forget: units. I stopped civ 6, because I was bored conquering civs with either only outdated or no units.

Having the AI spawn units from nothing is not nice, but where is the fun when there is no opposition?

With humankind at least I have a good story to tell after having played a game.

 

[Taefin]

The fact that the AI builds makers and research quarters goes a long way. A conquered AI city will often become one of my top producers of one or both. The decision to reduce their food (and gold?) upkeep by so much was a great idea IMHO as it simplifies the number of resources they need to manage. I couldn’t believe when I first discovered that AI cities in Civ 6 were procuring like 7 production at industrial after I captured them, with zero adjacency on all districts.

Also, Old World late game units looked like it was going pretty well lately, I haven’t played anything in a month or two now.

Oh and I can’t believe I missed all this mod.io action going on here! The better culture choice mod looks awesome, and stunning that Amplitude accidentally disqualified 70% of their cultures by making AI pick cultures with no new strategic resource requirements. Anyone know if changing AI build priories for more late game units in is in the works, or easy enough to figure out?

 

[PSG]

Having been lead play-tester on a fairly prominent title I can say I was truly shocked when the programmers plugged the Ai in a few days before we went Gold, but I wasn't shocked when the game underperformed saleswise. I don't think many if any games put serious thought and effort into the Ai.

 

[NombreyApellido]

it's usually a matter of studio resources. there isn't enough time, or budget or simply talent

i also worked in several games, mostly in the racing genre. back in those, the AI run in a very simplified physics engine. so it basically cheated and got the usual rubber banding help and all the tricks

but it was very subtle. if you were leading, it attempted a hail mary last corner pass and it could bump you hard but didn't wreak you (most of the time). or it nip you off pole position by a couple of tenths after a tense battle

the suspension of disbelief was never hurt

there's this old google talk from soren johnson in which he explains how cheating AI are a necessity in these very complex games but at the same time it is very important that they remain believable and feel fair to be enjoyable

 

[EscapedGoat]

Looking forward to the patch on the 28th, hope it has some balance improvements and fixes some of the dumb exploits which detract from the game like the "free city" exploit".

When it comes to the AI, it is dumb alright! But when was the last time someone praised a game for it's clever AI? believe i haven't seen that comment since Half-life (the first one)

A good example: I suspected that this game, as is the case with CIV6, the AI "picks" a city(outpost) tile when expanding. This is where it wants to build the outpost and where it will send the unit. If you then send a unit to block that tile, the unit that is going to establish the outpost just freezes and doesn't do anything (it could make the outpost on the tile next to it). In CIV6, it goes somewhere else after X amount of turns. Here it just stands there. The point of this anecdote (and exploit) is that this is not even really "AI" - it's just lazy programming and shows how low the ambition level is (in both CIV and HK). How hard would it be to add a statement: IF optimal target tile X = occupied, THEN establish outpost on Tile with coordinate X+1" (second best tile) or whatever. It's really not rocket science to make an "AI" that passes for relatively competent, but this is not high on the agenda of the developers, much more important that the game looks good on screenshots. Can't screenshot the AI!

 

[NombreyApellido]

there's also our bias at play... we might be a top percentile in terms of commitment as opposed to the huge majority of casual players that are fine with the challenges from a simplistic AI or prefer a mise en scene in which they are able to win playing "badly" or from a miraculous last minute recovery

i for one still can't get my mind around the idea that snowball economics are a big nono as if it wasn't a true reflection of actual human activity

to me, games can work as a learning tool and besides grand history, this genre is great for resource management basics

i mostly don't care if stone axes are made to be stronger than bronze spears. but it hugely annoys me the idea that growth is forced to stay linear by caps and cost bloating without a proper, canon, IN THE TEXT, explanation that make them tolerable

what does this have to do with AI? everything because if you are designing with an unrealistic objective in mind such as a linearly evolving economy, it will perform horribly against competent opposition of any kind

if linear is desirable, then code for resets after a given period and create the canon for them such as catastrophes and pest, or give rebellion back some relevance

i went back to Civ 5 to check my recollection of its happiness system as a regulator and had a blast playing against over expansion

the sad thing is an even nicer system is already present in Humankind but it isn't doing much

 

[Taefin]

I suppose to Amplitude’s credit, this simplicity of their oversights in AI so far does coincides with a studio rushing to meet a release date. Their current work rooting out pretty minimal bugs may be a promising sign that AI will be given some attention after the game is stable and some rebalancing done.

 

[KrikkitTwo]

there's also our bias at play... we might be a top percentile in terms of commitment as opposed to the huge majority of casual players that are fine with the challenges from a simplistic AI or prefer a mise en scene in which they are able to win playing "badly" or from a miraculous last minute recovery

i for one still can't get my mind around the idea that snowball economics are a big nono as if it wasn't a true reflection of actual human activity

to me, games can work as a learning tool and besides grand history, this genre is great for resource management basics

i mostly don't care if stone axes are made to be stronger than bronze spears. but it hugely annoys me the idea that growth is forced to stay linear by caps and cost bloating without a proper, canon, IN THE TEXT, explanation that make them tolerable

what does this have to do with AI? everything because if you are designing with an unrealistic objective in mind such as a linearly evolving economy, it will perform horribly against competent opposition of any kind

if linear is desirable, then code for resets after a given period and create the canon for them such as catastrophes and pest, or give rebellion back some relevance

i went back to Civ 5 to check my recollection of its happiness system as a regulator and had a blast playing against over expansion

the sad thing is an even nicer system is already present in Humankind but it isn't doing much

I think that is the key point...Snowball economics good reflection of real history...Snowball Politics (bigger empires having more stability) is not. And Humankind is full of Map painter bonuses.

Until they are willing to make empires collapse (which the Fame system could be great for if it was based on what you achieved in an era and not how much you Had (ie New population growth this era/New district built this era, etc.)... so you make a massive Classical Rome and scoop up the Fame, and then Collapse as the Byzantines... and get ready to earn Fame again as the Ottomans... and collapse as the Austro Hungarians, and get a decent mid level Contemporary Turks.... so you get 2 really high fame Eras, and you win. (no one else broke out more than once)

 

[PSG]

They like many other depend on multi-player to get around having to have a robust Ai. As far as developing a good Ai goes it's really not that complicated for a game like this. I laid out the basics of an Ai in this thread's opening post. I am not a whiz programmer by any means, but a few functions could handle the majority of the Ai's execution. If food <= X then build farm on best return food tile available else build MQ on best tile for production. Obviously there would be another 20 or so queries that would be part of the function such as foreign army strength within a certain radius, what phase it is, what culture it's playing etc... For goodness sake here were chess algorithms 30 years ago that are more in depth than today's games.

 

[NombreyApellido]

rome collapsed economically, it was a zombie empire long before barbarians and all that over simplified noise

a Roman legion (roughly 5,000 men) in the Late Republic might have carried into battle around 44,000kg (c. 48.5 tons) of iron – not counting pots, fittings, picks, shovels and other tools we know they used. That iron equipment in turn might represent the mining of around 541,200kg (c. 600 tons) of ore, smelted with 642,400kg (c. 710 tons) of charcoal, made from 4,620,000kg (c. 5,100 tons) of wood. Cutting the wood and making the charcoal alone, from our figures above, might represent something like (I am assuming our charcoal-burners are working in teams) 80,000 man-days of labor. For one legion.

source

but if size of a faction economy alone is a problem, then use DISTANCE TO CAPITAL as a regulator, ie increasing cost for far away everything and for desertion and mutiny

--

perhaps there aren't simple, bolts-on if thens

most current systems are layered behavioral trees and finite state machines. there are several easy to grasp content sources on the web such as youtube channels that go over these and there is also several academic white papers freely available on the specific subject of AI for turn based strategy games

there is also much work being done in very advanced experimental AI systems that are now able to beat pro level players at REAL TIME strategy games. those are of course completely outside the reach of any commercially available product but they are great to put into perspective the magnitude of "solving" the puzzle these games put forward

 

[Taefin]

I mean it gets pretty complicated when you move on to the sophistication of “If at war -> build units.”

 

[PSG]

For some reason I just remembered the old CIV Ai code that was "if player 1 turn away from building wonder then generate wonder"

 

[PSG]

I mean it gets pretty complicated when you move on to the sophistication of “If at war -> build units.”

LOL I don't think they even make it that far.

 

[reddishrecue]

Industry is so cool. Mass hunnic hordes can take out some of the best units on the game and because they're mobile.