Basic question re AI: RTS versus TBS

[Gedemon]

Complaints are virtually guaranteed. We get complaints now, would get complaints with the better model, and would still get complaints with variants in between.

A better question is what makes the game better, and how is that measured?

You can't scientifically measure that, it's a matter of taste.

For example immersion is mandatory for me, a better AI (at winning the game) that would break my immersion by playing in a non-RP way would make it worst.

And yes, it's also bad on that side...

 

[Trav'ling Canuck]

For example immersion is mandatory for me, a better AI (at winning the game) that would break my immersion by playing in a non-RP way would make it worst.

And yes, it's also bad on that side...

Strongly agree. As of RF, Civ 6 is a world inhabited by paper tigers whose behaviour doesn't vary in any meaningful way and who don't offer a meaningful challenge, rendering the diplomacy system both boring and irrelevant from my perspective.

Possibly that may change in GS. At least we saw some indication in one of Quill's videos that Cyrus behaved differently from the other AI in his voting. If that translates into more interesting interactions with the AI, I'd be happy.

 

[TheMeInTeam]

I'm not saying the AI trying a bit harder to win would flat out break immersion (though it could depending on how it's implemented), just that it's not what the Devs focused on when they made it. If you watch some of the live steams or interviews with Ed Beach it is clear that one of his favorite parts of the Civ experience is RP

He and his predecessors chose to do this at the expense of the "S" in TBS. There actually is a game genre that centers heavily on role play, and there's a reason that is the one with "RP" in it.

Development resources are limited and at the outset priorities have to be set. From what I can tell from the outside looking in one of the top priorities was to make each leader react to the player differently (preferably based on the historic attributes of the person the AI is impersonating).

I'm well aware that priorities need to be set. My assertion is that the Civ team has long set priorities poorly. In a turn-based strategy game, they cut corners on design of balance, performance, and user interface (!!) in favor of DLC content and role-play. Leaving aside that poor UI is strictly bad, these are strange choices to make in something marketed as a strategy title.

I personally think a bit more time could have been spent improving the AI before release

If one of your design choices is to make it weak on purpose by playing a different game than Civ 6, there is no point. They clearly chose that intentionally and still sold the game, so here we are.

You can't scientifically measure that, it's a matter of taste.

I didn't ask for a *scientific measure*. Since I didn't specify previously: it's a matter of stating what (for you or any particular person) makes the game more enjoyable and how to determine whether those element(s) are actually in the game.

It could even be something strange, like horse socks. If someone wants horse socks in their Civ game we can measure whether it delivers. If someone wants balanced victory conditions, we can similarly measure that this has been as prevalent in the Civ series as horse socks so far.

There should be *something* though. Otherwise you wind up catering to an undefined, incoherent taste.

For example immersion is mandatory for me

Define "immersion", in the context of your own preferences. What role play is okay for immersion, and what isn't?

The answer to this question is important, because it makes a big difference whether the stated preferences are self-consistent. When they aren't they should be rejected as a matter of course in game design, because there's no reason to slap mechanics into a game with incoherent reasoning.

The alternative is, of course, to create mechanics in a way where RP is not an active hindrance to winning. That Civ introduces RP behaviors from its competitors but neglects to do this is why I call it dishonest.

 

[ChocolateShake]

He and his predecessors chose to do this at the expense of the "S" in TBS. There actually is a game genre that centers heavily on role play, and there's a reason that is the one with "RP" in it.

I'm curious, has civ every been known as a hardcore strategy game? I've been playing since Civ 2, and I've never really felt that it was meant to be a game that forces you to play in a strategic way to win. There are some rules like not falling too behind in science and keeping a decent-sized army, but otherwise it's really flexible. Sure you can optimize, min-max, etc., and players who do that would win in multiplayer games - but in single player it's always felt like a game that you could relax while playing. That being said I don't play on deity - precisely for the reason that I enjoy being able to play the game as I like.

If I wanted a hardcore strategic experience I'd go to multiplayer games, I used to play Rise of Nations in a league, and that game was balanced amazingly well (you probably know it was Brian Reynolds who designed it). But then you'd have to follow meta strategies to win - there'd be no screwing around between equally balanced opponents. Games that are well-designed do allow for flexibility in developing new strategies though that can still be viable, though the vast majority of the time following a meta is needed.

It's all about the audience to me. If the game isn't meant to be hardcore strategy, then why should Firaxis invest in pumping up the AI? I'd say the AI is doing great as it is in fulfilling the expectations of the vast majority of the audience - except for when it had (and may still have) glaring weaknesses like being able to siege a city. That doesn't affect the vast majority of sales though.

I'd guess a lot of players are more interested in relaxing and building their own story, than facing a tough challenge each time they start up a game of civ. I know when I want to really feel challenged I'll load up a game of 1v1 StarCraft II.

Your question about what would make the game more enjoyable is intriguing to me. I'd say diplomacy is high up on the list for me. Lets get rid of things like "stop settling near me", stop X, stop Y etc. The game should instead focus on a negotiation system.

More creativity for victory conditions would be another one. They're slowly moving in this direction but hopefully they take bolder steps in 7 or the next expansion. With better diplomacy the need for a stronger AI becomes more important too.

 

[TheMeInTeam]

I'm curious, has civ every been known as a hardcore strategy game?

To my knowledge, Civ has never styled itself as "hardcore", and that's not the standard I'm applying. In asserting itself as a "turn-based strategy" game with rules and set victory conditions, there are some things that should be reasonably anticipated.

If I wanted a hardcore strategic experience I'd go to multiplayer games

This argument is a cop-out even when the devs use it. The AI is a sacrifice on the altar of poor design. If Civ worked well from a design perspective, the AI wouldn't have to be programmed not to try because the incentives would match dev-desired behaviors.

It's all about the audience to me. If the game isn't meant to be hardcore strategy, then why should Firaxis invest in pumping up the AI?

You're missing my point. The AI is a symptom, not the cause. At best, the devs use the AI in an attempt to cover major design flaws, knowing that most people *don't* play competitive MP in this game genre. Though that's partially because of its low quality in that format.

I'd guess a lot of players are more interested in relaxing and building their own story, than facing a tough challenge each time they start up a game of civ.

There are difficulty levels for that.

Your question about what would make the game more enjoyable is intriguing to me. I'd say diplomacy is high up on the list for me. Lets get rid of things like "stop settling near me", stop X, stop Y etc. The game should instead focus on a negotiation system.

Negotiation system isn't bad per se', but how does this look in PvP? Would MP players in a scenario with only one winner use it? Would you allow shared victories?

More creativity for victory conditions would be another one. They're slowly moving in this direction but hopefully they take bolder steps in 7 or the next expansion. With better diplomacy the need for a stronger AI becomes more important too.

Non-military victory conditions remain PVCs (pseudo victory conditions), because they are for the most part contingent on opponents not trying. When opponents do try, you wind up having to defend yourself militarily, or in the case of religious are effectively forced to win militarily anyway.

I do not call them PVCs as a matter of preference. I call them PVCs as a matter of the reality of in-game incentives in Civ 6. You can not and will not have a good AI while simultaneously forcing the AI to ignore the causal realities of winning in the game. It's impossible. AlphaGo can't beat me at Ultimate Chicken Horse. Why do players who want RP in TBS expect Civ 6 AI to be competent when it's actively instructed not to play Civ 6? That's silly.

 

[ChocolateShake]

Cheaper units for all or just the PC-players? I plan to modify some tile yields & combat strength _only for the human player_ to slow him down a bit, nearer to time when the PC-players start to manage victory.

Ah I've given it to all players - although I can churn out more units, the AI makes so many that I have to take any possible threats seriously, especially when I'm at war and the civs on my other borders have an alarming number of units, which they almost always do now. I play with the Timex3 mod too so the AIs get really serious about having huge armies.

Not a chance.

Civ6 is basically rock bottom of the series, Deity having direct bonuses to unit strength (never happened before), three settlers, and it's a joke. You can't even comprehend how much it will take to even remotely challenge good players at equal footing, let alone defeat a top player once at equal footing.

I was saying Firaxis could build a brutal, merciless AI. They clearly didn't go that route I think for many good reasons. @Rath_O_God highlighted the main ones.

Why do players who want RP in TBS expect Civ 6 AI to be competent when it's actively instructed not to play Civ 6? That's silly.

What is playing Civ 6 in your opinion?

I'm not claiming to have any answers, but the more I think about it the more it makes sense to me not to build an AI that actually poses a threat. A threatening AI would require a lot more effort and time and there really isn't a strong incentive to do so, since it would also require changes to the core gameplay of civ so that most players could enjoy it.

I think I might be understanding your point more now, am I on the right track? I would like a game that has a threatening AI, but I don't have any expectations for that to be civ. I just want Gilgamesh to protect his spaceport a bit more and for the AI to occasionally take a city when I really am not paying attention to my army, so I either have to build more units, give them concessions/tribute, or have to go down a level. As long as there aren't any glaring weaknesses, I'm pretty satisfied with how things are with civ, but that's just the way I play it.

 

[Rath_O_God]

He and his predecessors chose to do this at the expense of the "S" in TBS. There actually is a game genre that centers heavily on role play, and there's a reason that is the one with "RP" in it.

Developers mix and match genres all the time and some times it works, some times it falls flat. There is no rule saying you can't have RP in a strategy game. Often times when Devs mix and match you get one genre that out shadows the other, or one is heavily diluted. And you could defiantly make the augment that Civ has spent too much focus on the RP to the detriment of of the TBS aspects (as well as balance, but I feel like that would fit better in a separate discussion)

I'm well aware that priorities need to be set. My assertion is that the Civ team has long set priorities poorly. In a turn-based strategy game, they cut corners on design of balance, performance, and user interface (!!) in favor of DLC content and role-play. Leaving aside that poor UI is strictly bad, these are strange choices to make in something marketed as a strategy title.

Trust me, I feel time management prior to release could have been much better spent. The Firaxis team are not the only ones who "cut corners" throughout development. Its a sad reality that came with the implementation of the "We'll just patch it later" mentality and I swear it gets worse by the year. We could go back and forth for days on who the worst offender is, but Faraxis at the very least is among their ranks. That's why I don't normally pre-order (even for things like GS which looks fairly good from the pre-release stuff). Its almost enough to make me want to just stick with the games I have until the industry can get its life together and release balanced, complete, and working games at release again.

If one of your design choices is to make it weak on purpose by playing a different game than Civ 6, there is no point. They clearly chose that intentionally and still sold the game, so here we are.

Just because some people (myself included) want an AI that RPs a bit throughout the game does not mean we want it completely incompetent. It not having a basic understanding of several core mechanics is just as immersion breaking as it going all out and steamrolling me in the first 50 turns. this goes back to time management and the "patch it later" "fix it in post" mentality that is so abundant in major games nowadays. If it is halfway serviceable and can look pretty for the first 3 hours of play it's good to ship.

I'll close with this. I think the only thing we really disagree on is what is more important in the single player mode. And your opinion on what makes the game "fun" is just as valid as mine or anyone else. If enough of the fan-base lines up with what you want out of a Civ game I hope the Devs take it to heart and shift priorities in the future. That is the only way the series is going to develop and grow. Devs who don't listen to fans will eventually find them leaving for other titles. And even a cursory glance at any of the Civ forums will reveal that a substantial proportion want an AI that can at least hold its own on the upper difficulties without all the buffs. They want it to play more logically, and be a challenge because of its decisions, not its stat changes.

 

[TheMeInTeam]

What is playing Civ 6 in your opinion?

Making decisions in pursuit of one of the game's stated victory conditions. This includes but is not limited to attempting to prevent your opponents from winning also.

I'm not claiming to have any answers, but the more I think about it the more it makes sense to me not to build an AI that actually poses a threat. A threatening AI would require a lot more effort and time and there really isn't a strong incentive to do so, since it would also require changes to the core gameplay of civ so that most players could enjoy it.

Saying a threatening AI "isn't fun" is disingenuous by developers. It isn't "fun" because the incentives in the game dictate players do things that wouldn't be "fun".

But those incentives are part of the design. If the devs are making a case that the AI optimizing within their mechanics isn't fun, they are making a case that the game itself is poorly designed.

So you're right in that sense. There are changes that should be made to core gameplay, rather than making a game and then having the AI not play it because even the devs themselves know it would be bad in that case.

I think I might be understanding your point more now, am I on the right track? I would like a game that has a threatening AI

I want a game where there is actually incentive built into it such that its mechanics are usable when everyone is trying. This means that diplomacy is more significant in MP, for example. The design in Civ games is more flawed than the AI in Civ games, and you can't meaningfully fix the AI unless that changes (I suppose you could, if you want to make them all ruthless warmongers, but I don't anticipate that being popular overall).

There is no rule saying you can't have RP in a strategy game.

True, but there *is* an expectation that strategy is not actively undermined in a strategy game. WRT balance, it's worth mentioning here. If the game isn't balanced but you instruct the AI to pretend it is balanced, you get poor AI performance. Civ 4/5/6 all did this to their detriment so it's not a good thing to brush off.

Trust me, I feel time management prior to release could have been much better spent. The Firaxis team are not the only ones who "cut corners" throughout development. Its a sad reality that came with the implementation of the "We'll just patch it later" mentality and I swear it gets worse by the year.

Sadly, the market allows it. They'll even throw money out there before the product is finished, then play the game at retail while it still isn't finished per standards of good games. Not everybody does this of course. Some devs are open about early access status and even have relatively polished games by the time they swap it to 1.0. Mostly relegated to the good examples from the indy scene, but it still happens.

Just because some people (myself included) want an AI that RPs a bit throughout the game does not mean we want it completely incompetent. It not having a basic understanding of several core mechanics is just as immersion breaking as it going all out and steamrolling me in the first 50 turns.

The problem is "a bit". What constitutes "role play" vs "immersion breaking" rarely has a coherent definition. Even to the people attempting to define it, it's a struggle to be self-consistent in stating how much "role play" is sufficient.

That misses the point anyway. The problem with Civ 6 isn't "role play", it's the mis-matched nonsense setup of alleging allowing for "role play"/"different victory conditions"/"unique agendas" and then turning around and providing incentives per the rules that do not align with those whatsoever. In proper, well-designed role playing games the player has INCENTIVE to role play in some capacity, not active disincentive!

Civ could go that route, opting for role play and eschewing its present win conditions/war incentives/4x roots. I wouldn't prefer it, but I would respect it a lot more than the constant misalignment of incentives we've seen over the past decade+.

 

[drubell]

I suppose an argument here would be that Firaxis, reading the room, chose not to design AI that was meant to stop you, but rather one designed to challenge/pester/bother you to create an artificial form of "conflict." An AI designed to stop you would infer that the majority of Civ players play to see the end of the "match" which I don't think is what their statistics read. I read from these forums often that players who play to win stop playing when it's clear they will win the game in 50 turns or whatever. We hear even from the "roleplayer" crowd that at a certain point in their games, they stop once they finish their own pseudo-victory.

Length of expected play in a game is also something to note.

Something I notice is that almost every key system in Civ 6 shows up very early in the game at around Classical. Heck, people are quirking their eyebrow about World Congress showing up at Medieval, but it makes sense if Firaxis reviews their user statistics and sees that the majority of players stop their games at around Renaissance or Industrial. If people aren't playing Civilization to complete the games, then it doesn't make sense to focus the AI on trying to complete the game either.

A Starcraft 2 match is generally about 20 minutes at longest, no? Then yes, players should expect the AI to be built around a satisfying and challenging endgame. Civ games are not 20 minutes long (and turn times are a problem but you get the point).

If Civ 6 was focused on the endgame then I don't think we'd still be in the dark 2.5 years later on any documentation in the game about how to generate Domestic Tourists, for example. Because that kind of information is late-game stuff and therefore isn't priority to Firaxis.

Now I don't advocate a "play to lose" AI. That's part of why I've still chosen not to purchase Gathering Storms. I'm just saying that if Firaxis recognizes that its players don't play to finish games, then it wouldn't make sense to design the AI around moving toward win conditions ASAP.

 

[TheMeInTeam]

A Starcraft 2 match is generally about 20 minutes at longest, no?

They sometimes go longer. I've seen best-of 5 series that conclude after 4 games and go ~2 hours. However one game was over in less than ten minutes, so the other 3 took the remaining 110 minutes. Still far shorter than Civ on a per-game basis.

Given what you're saying, it's strange that Firaxis also chose to put considerable resources into the future era.

Also, there's a reason players often don't finish civ games, and this reason is not consistently shared by good games. It doesn't have much to do with the AI, but you're right that so long as Civ remains in this state altering the AI won't be very impactful.

 

[cvb]

Ah I've given it to all players - although I can churn out more units, the AI makes so many that I have to take any possible threats seriously, especially when I'm at war and the civs on my other borders have an alarming number of units, which they almost always do now.

Sounds good! Can you please give roughly the factor of your modded units being cheaper? 2, 5 or even a magnitude??


I had again a look into the "AI Victory Times" Test (passive human player), which Trav'ling Canuck did on Deity Level: T245 prior to AI victory, Frederick Surprise War: Player Eliminated / T351 Science Genghis / T307 Science John Curtin / T318 Science Poundmaker / T310 Culture Teddy Roosevelt / T330 Science Pericles / T299 Science Shaka / T301 Science Lautaro / T327 Science Montezuma / T280 Science Robert the Bruce / T319 Science Gilgamesh / T323 Science Poundmaker/Trajan / T340 Science Gorgo / T327 Science Jadwiga / T321 Science Hojo I think, the results are quite close together - ie. unsymmetric modifications could move the human player into more interesting territory ...

 

[ChocolateShake]

Sounds good! Can you please give roughly the factor of your modded units being cheaper? 2, 5 or even a magnitude??

Ah yup at the moment units are at half price, I think I played around with much cheaper than that too as you said I also went for enabling corps and armies from the start of the game too earlier, and with giving melee units a supercharged version of Macedon's UU district bonus, but I rolled those back as I've been testing the cheaper units. I definitely think your idea of only giving the bonuses to the AI is great!

I think, the results are quite close together - ie. unsymmetric modifications could move the human player into more interesting territory ...

Awesome! Looking forward to it

 

[phoinix]

If I may add my 2 cents:

A lot of people say that if they improve the AI too much it will ruin roleplaying and civ is partly a roleplaying game. I agree. However, there are situations where the AI could be improved a lot without ruining anything and that is in the tactical section. It's really not fun or immersive seeing AI doing stupid mistakes or failing to play the game properly.

Also regarding the "too hard will brake immersion" argument: I don't see it as black and white like that. Can't AI be just hard enough? In recent research, it has been shown time and time again that performance of an ai at playing a game increases with training time (epochs). So couldn't they just simply stop training and freeze the AI when it comes close to human level performance? Different difficulty levels then are just the same AI trained for different times. No bonuses or rule changes are required.

Also when changing the game rules somewhat, there is no need to re train the AIs from scratch. There is a method called transfer learning where we use a pre trained model to reach good results with much less data. Same can be applied here. Firaxis trains the base models (one per civilization, leader) and then slightly retrains them whenever a game rule is changed.

A lot of people are also saying that ai is very hard to improve for civ because its rules are too complex. I believe that it's just a matter of having well defined game rules. There are numerous examples now of AIs besting top players in complex games. I don't see how civ is different. It's also a game defined by a set of game rules. A similar methodology should apply.

 

[TheMeInTeam]

Also regarding the "too hard will brake immersion" argument: I don't see it as black and white like that. Can't AI be just hard enough?

No, it can't. "Hard enough" for one player isn't the same as for another, or even the same for one person over time in many cases.

A lot of people are also saying that ai is very hard to improve for civ because its rules are too complex. I believe that it's just a matter of having well defined game rules. There are numerous examples now of AIs besting top players in complex games. I don't see how civ is different. It's also a game defined by a set of game rules. A similar methodology should apply.

As I've pointed out, the developers set game rules and incentives that encourage certain choices. Then they concluded it wouldn't be fun for the AI to make these choices.

There's no reconciling that. As you say, the game rules need to work for the kind of game they want to be played first. Civ devs haven't accomplished that yet, and until they do they will continue to make the AI bad deliberately on top of its typical tactical failings.

 

[phoinix]

No, it can't. "Hard enough" for one player isn't the same as for another, or even the same for one person over time in many cases.

But that is why we have difficulty levels, so that each player can customize their challenge.

As I've pointed out, the developers set game rules and incentives that encourage certain choices. Then they concluded it wouldn't be fun for the AI to make these choices.

There's no reconciling that. As you say, the game rules need to work for the kind of game they want to be played first. Civ devs haven't accomplished that yet, and until they do they will continue to make the AI bad deliberately on top of its typical tactical failings.

Civ AI could be a series of smaller AIs, hierarchically arranged. Tactical AI is one of them and the thing with tactical AI is that it can be treated in isolation.
In order to maximize the returns of a battle you don't need that much info really: you need information on unit movement, unit strength and terrain. The tactical AI needs of course to be able to optimize for all combinations of these parameters. That is why there is no need to finalize all the game rules before developing it. Only a very small core subset of rules would be enough (we use hexes, we have passable and impassable terrain etc)

In a battle, you need to optimize the results of all your deployed units at once. That sounds a lot like chess. The differences are that: a) the board (terrain now) is not uniform b) units have different strengths (but the same logic as in chess, of trap setting and covering each unit with others still applies) c) units may have secondary effects that are irrelevant to the outcome of the battle itself but help in the overall strategy.
Chess has been solved. I don't see why a civ battle cannot be treated similarly. The only reason I can think of is that turn times would suffer.

Still, an improvement to the tactical AI would improve the overall AI effectiveness immensely imo, because let's face it, war is kind of an exploit right now.

 

[TheMeInTeam]

Civ AI could be a series of smaller AIs, hierarchically arranged. Tactical AI is one of them and the thing with tactical AI is that it can be treated in isolation.

I thought it did something like this already. Seemed to be the case per a YouTuber's case in Civ 5. It's not like the AI in civ games was ever a cohesive singular thought entity. It's just easier to discuss by clustering it that way conceptually when talking about it.

I did suggest something similar earlier in this thread, to run these simulations only for tactical AI, since this would have fewer possible factors than the game as a whole by a huge margin.

Tactical AI improvement is one of the few things that would be unambiguously good for the game, since even the role play crowd seems to agree that its performance in this regard takes away from the game.

 

[Cakeathon]

As for OP's point, the in-game AI is not the same as AlphaStar. Also, I don't think we need a super-human AI per se, we just want a decent one.

About resources:
I will dissent with people claiming the problem is ressources, I think you can make a very good AI without it necessarily being at the level of what deepmind does. The reason Starcraft 2 has good AIs available (not just alphastar) is because they have an API for machine learning that allows programmatic interaction with the game. This means anyone can try to create anAI for it, they even have AI tournaments where you have teams from different schools competing. You can even literally download an AI somebody created as a mod and play against it. That's how you get good AI for cheap, you just crowdsource it.

About lower difficulties:
Some people say you can stop early for easier AI, that is true if all you want is an "easier" AI(in that it wins less often) but it's not really true if you want an AI that behaves in a manner that seems coherent for a human which I think is what people would want. People have the impression that AIs progress must be similar to humans, but it's really not. They will rapidly become super-human at certain tasks and look like complete idiots at others. Only when they reach the top do they actually start to actually make sense to a human since they have ironed out all the nonsensical decisions that hindered them at previous iterations. Even alphastar in the last game couldn't deal with Mana moving in and out of the fog, something even low-level players would be able to deal with and yet I'm sure even the worst versions of alphastar would crush any human at stalker micro despite building their expansion in the middle of nowhere.

There are better ways to accomplish this but they are model-specific. For example, alphaGo could be made worse by limiting its time or processing resources since it still uses a tree search algorithm at it's core, it is just improved using neural networks. But other AIs are just a simple feedforward so you would have to find other ways.

Pro player data:
Training on pro-player data is not mandatory, it just makes things easier at the beginning. For example, AlphaGo was trained on a database of human moves but it's successor AlphaGo Zero wasn't.

Constraining to tactical AI:
I can definitely agree with this and it also would allow the AIs to keep the personalities and so on. It should actually be fairly easy to do as well since it's a very constrained problem, that's the kind of stuff you would do as a project in an AI class. Random maps/units and asymmetry are not nearly as big a problem as some people in this thread claim when the problem is so constrained. But it's harder to crowdsource such a limited part of the game so they would probably need to hire somebody to do it.

 

[cvb]

Want just mention an article in the current (5'19) ct magazine (german): Wie die DeepMind-KI AlphaStar Profispieler in StarCraft 2 besiegt hat

StarCraft-API PySC2 (Repository & Blogposts for Open Source @ GitHub)

 

[God of Kings]

I took another look and I found that there is much more for AlphaStar before it can be the best StarCraft II player.

It can only consistently win in a Protoss vs. Protoss match on a single map with something akin to a maphack in favour of the AI. At least that is a good first step.

It would be like the AI only playing as a pure vanilla civ vs. a human player playing as a vanilla civ on a pre-made map with only domination victory enabled.

 

[cvb]

At least that is a good first step.

And the rest is just a matter of time ... probably much less time then some think - I mean, if I drive today a bus or train, then feels the idea to land on the moon without a single microprocessor very, very unreal ...

Civ6 could benefit a lot from such at least in tactical combat: comparably few units, small area, limited reinforcements ... we all know those stupid moves where no bonus to the AIplayer would change anything.