Civilization VI


Real AI techniques – for example the tweaked deep learning approach used by alphago – have virtually no bearing on the so-called “AI” in a game like Civ VI.

In Civ VI, the challenge is encoding ordinary high-level logic, often in a simple rule-based form, in combination with simple algorithms for pathing and numerical calculation. This is not an intellectually difficult problem; it’s just a question of investing time and resources in getting it to work. The Civ series has historically laughed at this investment and done the minimal work necessary to give the computer nations the appearance of some kind of sensible play.

Unfortunately, it seems the designers in Civ V and especially Civ VI have failed to understand that adding decision-making complexity and burden to the human player also adds it to the computer player. Or if they do understand, they don’t care. So they keep coming out with game systems that they aren’t prepared to code properly for the computer player. They weren’t prepared even to implement competent pathing for one-unit-per-hex back in Civ V, and now with every expansion adding yet more stuff they don’t really understand, the computer players appear more and more incompetent.


If you read the Nature article on AlphaZero you will see that there were a few pieces of knowledge in there:

  • The rules of the game were coded by hand
  • The NN was designed by hand
  • A subtantial number of hyperparameters were adjusted by hand

Great achievement, yes, but not magical. As for the applicability of the technique, we will see, it is far from clear that we get our money back on the computation time required to train the neural networks that essentially learn a heuristic to play with, or to evaluate the moves.

On the gap between Go and Civ, let me make you a list:

  • Initial game states are different from session to session
  • The board is not fully observable
  • The board positions have multiple attributes, several change over time
  • The number of moves possible depends on number of cities, units and technologies
  • The number of units and cities changes over time
  • There are several adversaries
  • Adversaries may or may not be trying to achieve the same victory condition as you
  • Adversaries may switch between those during the game
  • Outcomes of moves depend on chance

2030 was optimistic… more like 2080 perhaps. Let’s all play just Go until then :)


I think they simply do not care. The AI can be done. AoW 3 has a fantastic tactical AI and a decent strategic AI. Also, Civ 5 had a shit AI when it released, and eventually modders put in a real AI to make up for it. Firaxis will simply does not care enough to pay an AI programmer to solve their problems.

I will not be buying another Civ game until I get a green light from other players about the AI not being complete shit.


Sorry, search isn’t finding a Civ V thread.
Grabbed Civ V on steam sale, can someone give a quick rundown of the best mods? AI being the most important as the lack of AI is why I’ve never moved past Civ 4.

Vox Populi was mentioned earlier in this thread, would that be all I need?


There’s a bunch instead a singular thread, as Civ 5 has been one of the more contentious games in QT3. I have more hours in the game than most people (currently over 4000). Still, my opinions are just that as I’ll try to explain in a bit.

That’s a great AI mod.

Quality of life mod:
Infoaddict - adds a vast amount of perusable data via a new option in the drop-down. You won’t need to wait for “Who has the pointiest stick” reports nor deal with insipid advisers to get useful insight into how your civ is doing compared to others.

Suggested enhancements:
Religion: Beliefs Expansion Pack - adds many more religion options to give a fresh coat of paint)
Religion: Improved Customization - want more? Okay, every nation gets the Byzantine-style bonus belief. In exchange for this not being unique to them, the Byzantines get free shrines.
R.E.D. (Regimental & Ethnic Diversity) mod set - makes units look more like armies and adds civ-specific touches to differentiate
City States Leaders for G+K (Flags) - get to see the flags of various city states instead of the generic screen when talking with them. Nice for history/flag fans :)

Civ 5 has more mods than you can shake a stick at, essentially allowing you to “fix” almost anything you think needs fixing. The AI is the only weak spot, but Vox Populi goes a long ways toward helping. Still, keep in mind Civ 5 isn’t really designed for gamers who demand all sides play by the same rules and yet provide an equal challenge to the human.

edit - finally, I’ve posted this mod before and will do so again:

In the “just for the heck of it” category, a very simple little mod that my kid loves which can no longer be found on the workshop:
XML file

<?xml version="1.0" encoding="UTF-8"?>
<!-- Created by ModBuddy on 10/22/2010 4:41:38 AM -->
- <GameData> 
  - <BuildingClasses> 
      - <Update> 
               <Set MaxPlayerInstances="1"/> 
               <Where MaxGlobalInstances="1"/> 
      - <Update> 
               <Set MaxGlobalInstances="-1"/> 
               <Where MaxPlayerInstances="1" MaxGlobalInstances="1"/> 


<?xml version="1.0" encoding="utf-8"?>
<Mod id="3808993b-e543-47de-abac-c849816269b3" version="1">
    <Description>Wonders are converted to National Wonders.

Any civ may complete any wonder without regard to what other civilizations do.</Description>
    <SpecialThanks>Firaxis and Kael</SpecialThanks>
  <Dependencies />
  <References />
  <Blocks />
    <File md5="EF9DD81B7DA0CE05251855B904ABA989">XML/Buildings/CIV5BuildingClasses.xml</File>

The above mod makes it so every World Wonder becomes a National Wonder, allowing every player to go Wonder-Crazy (and yes, the AI goes to town as well). Keep in mind, there are a couple of policies which provide benefits to cities with World Wonders, and these become useless without further modding. Likewise, there’s one that gives bonuses for National Wonders, and this becomes overpowered without further modding.

It was once called “jmknpk_Wonders_All_NationalWonders(v1)” in case you want to scour the internet for it. Its author (jmknpk) no longer appears to be active.


Dang! Thanks for the quick and in depth reply. With Civ I’m more for the story of my civ than min maxing to win at any cost so as long as the computer cheats aren’t totally obvious I can overlook them.


Sounds like you’ll enjoy the game, then :)


As someone else who also doesn’t hate Civ VI, I hope you don’t mind me answering your question.

Personally I enjoy Civ games for exploring and understanding how the game systems work. Not to the extent of min-maxing everything to a 150-turn spaceship victory or whatever, but rather in looking at the game design, considering what the designers were probably trying to achieve by a particular decision, what works and what doesn’t, what emergent strategies there are. This is interesting (to me) for most games where there’s not one right path to victory but multiple competing requirements, the best of which is dependent on the particular game situation.

So coming from that perspective, I’d say to take a look at the way the game systems intermesh, particularly the ones brand new to the game rather than inherited from Civ V.

For example, the research boosts system (that Tom disliked so much). I feel it produces a number of interesting effects, some good, some bad. The foremost of these is the way it acts as a brake on the progress of the research leader (without feeling punitive) by making them spend extra science / culture if they can’t achieve boosts to match their science rate. In turn this provides a pleasant positive reinforcement to the player who can manage this: when they have to work hard to get particular boosts and they manage to line them up at just the right time for the research.

On the minus side, it can act to limit available strategies by making any that don’t chase boosts inefficient. Also, some of the individual boosts are poorly designed: either too random, trivial to achieve or, worse, force you to do actions that provide no other bonus to your civilization and are just a production sink. And of course the AI can’t handle the boosts effectively, so in single player they act as a huge bonus to the player alone.

In a similar vein, here’s a rather longer description of the knock-on effects of the One Unit Per Tile system (written for Civ V, but still largely appropriate). Credit for this goes to Luddite from the Realms Beyond forums.

Hidden due to length

They wanted to keep the “civ” feel to the game, where you settle new cities, build improvements and city buildings, and go in to the city screen to adjust your citizens. This meant that they had to limit the total number of tiles in the game, and so they tried to force army sizes to be very small [to avoid traffic jams]. A typical civ 4 army of ~50 units would be incredibly annoying to manage in the Civ V style, so they wanted to encourage armies of only 5-10 units.

In order to do that, they had to limit production. You can see that in the decreased yields: production and food yield have been decreased compared to civ 4, whereas the food required to grow a city was greatly increased. The early units like warriors don’t take very long to build, but the cost of units quickly increases. The high upkeep costs for units, buildings, and roads factors in to this as well. The idea was, I think, that every new military unit would take about 10~20 turns to build, just enough to replace your losses while you continually upgraded your original army. As a result, your army size would stay almost constant throughout the game.

So now the developers are stuck with a game that has greatly reduced production values. That’s fine, except for one thing: what do they do in the early game? They can’t expect us to just sit around clicking “next turn” for 40 turns waiting for our worker to finish, or 100 turns for a library to finish. It’s bad enough that it already takes up to 15 turns to finish that first worker. So, they had to make the early stuff a bit cheaper. You can build a warrior in ~6 turns, and you can build a horseman or a library in ~10. Even a coloseum only takes ~20. The idea was that a small city was efficient enough to produce the early game stuff in a reasonable amount of time, and as the city grew, it would produce the later stuff in the same amount of time - keeping army size constant while the cities grew and built infrastructure. There would be no massive increases in the power of a city with its size (like civ 4 had) because if a city became really powerful, it could create huge armies which would break the 1UPT system. If large cities were only modestly more powerful than small cities, the army sizes would stay small.

What the developers overlooked was that we’re not limited to just a few large cities- we can build as many small cities as we want! Granted, we’re limited a bit by happiness, but there’s a lot of ways to solve that little problem (like keeping the city size small). And since small cities are so efficient at building the early game stuff, and large cities never become vastly more powerful, the many small cities with their trading posts (even without any multipliers) will quickly outproduce the large cities with their mines, despite their forges and workshops.

The game is in an awkward situation where large cities can’t be too good because it would imbalance the middle and late game, but small cities have to be good or else the early game would be boring.

==End of hidden text==

Anyway, I hope that gives you some ideas. I’d also recommend playing multiplayer if you can find committed people of an appropriate skill level to play against (and ideally someone to make a balanced map). So much of what’s wrong with Civ VI lies in the AI; against competent opponents most of the design holds up well, providing interesting decisions and satisfying combat. Though it does start to fall apart later in the game, after the renaissance or so.


Love your in-depth response. It’s really nice. Thanks!


It sounds like your sister is not nearly as serious a gamer as you are. What is her opinion of Civ IV, V, vs VI?

One of the problem of forum of hard core gamers is we have slightly warped view of what makes a good game.

It is not like a only like complicated games, I found both Candy Crush, and Angry Birds pretty addictive.


Civ IV, she and I played extensively together for years. She played a lot more single-player than I did, and after I don’t know around 5 or 6 years I said I needed something different.

Civ V, she purchased probably within months of release, enjoyed and pushed and pushed and pushed me to play, but I declined and never bough itt, but we had other strategy games going at the time.

Civ VI she purchased I think with 6 months of release or so again, and she pushed me to get it at a sale, which I did. I will play it when she asks me to.

I would say I am more of a hardcore gamer than both my sisters, but this sister, she played Dwarf Fortress for a couple of years, basically until she had her baby and decided the juice wasn’t worth the squeeze anymore. During our MMO days, she was in a top guild and raider as a healer, like the healer, and I just sort of joined their guild for various adventures, never raided really. Her significant other called Thea the relationship killer once because not only does she and I play that a lot, she excitedly calls me anytime there is a new newsletter out for Thea 2 because she knows I don’t subscribe to their newsletter. I got her into Thea though once I found it added MP.

So her seriousness is more related to her interest level than anything else. Her temperament is also different than mine, doesn’t get as frustrated or irritated at a number of things that drives me nuts. So when some AI declares war at me and can’t land troops, I’m like, this AI can’t play the game. They declared war on me and doesn’t even have ships to land, wth. My sister’s response is sucks to be them, you’re going to wipe the floor with them right, and then she’ll start talking about whatever wonder she is going for.

If you can get her into a game, she will really, really get into it and do all the homework for me. On the other hand, I explore more. I buy and try more games by far. My Steam library dwarves hers and it’s miniscule compared to what I’ve seen others say they have here.

She loves Civ. Just the idea of it, the way you play from start to finish, the historical pieces of it and the unique units, wonders, like the whole package. The only Civ game I think she has ever regretted purchasing was maybe one of the Call to Power games, and that’s because we didn’t have the same resources then to know about that weirdo split.

Apparently she played Fallout 76 not too long ago, her so invited her. She called it a buggy mess. She’s tired of zombies and those are zombies no matter what they call them, and yeah, not happening. I told her so we’re lucky she plays 7 days to die at all. He’s pushing it! She would not play that game at all without us. Monster Hunter she might have but anytime my nephew shows up during our play sessions, she winds up dying. Not an issue with CIv either.


Basically alphagozero just know the rules. It doesn’t use human knowledge of the game at all.
Then in a few days by training he can go ahead of what humans researched for thousands of years and even beat them

CivVI is a different game, different rules doesn’t mean it’s so complicated. First you can dismantle in sets which can use different way of AIs (like heuristic) not only the last updated from Nature. To me this game doesn’t seem to have a combinatoric as deep as Go. (Number of atoms in the universe, ya know that I guess) look at chess: unlike Go, you have different pieces, they move… That doesn’t make chess so complex compared to go

I may be wrong ofc

Coming back on the civ series, I did enjoy a civ V well modded (with simfanatic offers if my memory is good)They made marvels combining civ IV in it. So I push everyone to try it.


I’m somewhere in the middle on how effective that sort of machine learning would be for Civ VI. I think the bar is so low it couldn’t fail to produce a much stronger AI than the current one, though. :)

It’s worth noting that it wouldn’t take just a few days to get there. Playing against itself, it was able to play hundreds or thousands of games of go a day. Civ VI isn’t as well optimised for playing at that sort of speed.

Also, “just knowing the rules” wouldn’t be as simple. There are all sorts of bugs, different ways of rounding, minutiae of resolution order, that sort of thing that mean I’m pretty sure even the devs couldn’t write down the rules of the game without a significant amount of effort. You’d certainly end up with an AI that exploited every bug to the max.


I gave you a very concrete list of differences which aren’t trivial. I also listed three things that were specified by hand, which you only learn about if you read the actual papers not the promotional blog kool-aid.

You just did some jazz hands at me.

What you said @rho21. A big issue using Monte Carlo Tree Search is the sheer computational cost of constructing the tree of game positions and moves putting those together. It took 40 days to train AlphaZero and that on a custom made super computer of which as far as I know there’s exactly one in service (Deepmid’s).

OpenAI had an, arguably, easier time with DOTA, accessing the game state via a programming interface, and constructing a simplified yet informative representation of game statwz. That was used by the AI players that got hyped a few months back. But OpenAI weren’t doing as Deepmind’s AphaZero, that plans moves ahead, they were just training a NN that predicted the payoffs of the games moves.

Regarding fitting to bugs etc. That has happened already and quite often. Deepmind Atari player broke the game of QBert, exploiting a bug that lied dormant for 30 years.

You may be interested in the, quite amusing, examples compiled on this wonderful blog


Back to Civ VI are there any must-have DLC races/scenarios to get?


I did read carefully your concrete list and did answer giving you 1 famous example of something which looked more complex (chess) but which is not.
To say that the Nature paper is a lie ( by omission) sounds a bit strange in a research community. They even got a AI prize I think for their work? I did just report what they wrote: no human knowledge on the game. Is it a promotional lie? In fact I am really interested in that concern so if you could explain a bit more your opinion, I would read it carefully too.
Please don’t take offense as this is absolutely not my intention.

I myself was one of those people who would never believe it could happen or at least not before some decades. And the second shock was when they made this “no human knowledge” version which is even stronger.

Concerning the use of a super computer, that’s true but now you can run on your own PC (better with a good GPU) a lighter version which is still quite good. Alphagozero did enter everyday life of many go players (even professionals of the game) as go assistant. And that could interest the civ players of the future.

There are interesting points in the debate around this AI aspect I found, like to make the game more compatible to the AI as the AI to the game, or how the AI would maybe first more debug the game mechanics. I’m far away to think it would be a task too complex to not enjoy it in our civ games, because you know I am getting cautious on this kind of prediction in which I went wrong twice already.


Chess was “solved” 20 years ago - remember Deep Blue vs Garry Kasparov? Not sure yet what was the point of that. Especially since chess hasn’t any of the features I listed.

Who says that? The paper is fine. You can find all the nitty gritty details there.

The blog post you linked is a different matter. Is hyping the same as lying? Americans have the expression “astroturfing” for that.


I listened again to Soren Johnson’s Designer Notes episodes with Brian Reynolds and @Sid_Meier (dang, that @ is the only one that wouldn’t work), and they discuss regularly the city spam issue of earlier Civs and the long road to mitigate it… I don’t know if they ever succeeded — I am anything but a Civ power player — but it seems, reading you, that Civ 6 re-opened Pandora’s box!

Pardon my useless venting, but I need to type it: Civ 6’s AI is so infuriatingly bad it turned pacifist-baby-culture-victory-me into warmonger-bloodthirsty-new-me then, and in all the strategy games I have played since. I feel I am still angry at it.


Trying to balance Tall versus Wide empires is always an interesting challenge since ideally both would be viable strategies in the game but it certainly doesn’t always work out that way. From what I remember of Civilization V multiplayer the power play was to work your way up to four well developed cities. So Civilization V leaned towards the other end of the spectrum and typically universally favoured taller empires over wider empires.

Haven’t sunk enough hours into Civilization VI to know much about what strategies are considered optimal and such. Wouldn’t surprise me if Firaxis accidentally skewed things back towards wider empires being the preferred strategy in response to Civilization V.


Wide was the way to go in 5 too.

That guy did a ton of good Civ 5/6 articles. His latest one he decided to test just how broken the Civ 6 AI was by trying a challenge game.

-Deity AI
-Cannot build or own districts
-Must win by peaceful space race victory
-No declaring war on or capturing AI cities

So a pure economic game with a huge economy handicap against the highest difficulty level. Guess how it went?