Do you want an AI that can make reasoned choices?

[shevvek]

People want an AI that's fun to play against and that poses a moderate challenge but doesn't win the game. People want an AI that does things to remind the player that it's playing the same game.

This, exactly, is what I want, even as someone who routinely plays on Deity. My issue with the Deity AI is that it poses basically zero challenge after the early game, and that it does immersion breaking silly things like not build a Government Plaza until turn 200. If a difficulty setting were added where the AI uses every aggressive dirty trick, I would probably play a level below that. If I wanted the experience of playing against a ruthless enemy using every possible tactic to win, I would play against human opponents.

 

[Leyrann]

(Too soon to have an endgame scenario about defeating a global pandemic?)

I think it'd be disingenuous to have a modern-day pandemic that's more impactful to the game than a medieval era pandemic. To put things into perspective, the black death killed five to ten times more people than the estimates for covid-19 (themselves three times higher than the reported number) despite the world population being sixteen times lower at the start of the pandemic than it is today. That's over a hundred times more deadly compared to the total population.

Even if you take a lower estimate measure of the bubonic plague's mortality rate and a higher estimate measure of untreated covid-19 mortality rate in the early days (so not omikron), you still get a ten times higher mortality rate for the bubonic plague, and it was the least deadly (albeit most common) of the three forms of plague.

Diseases belong in the medieval, renaissance and maybe industrial era.

 

[kaspergm]

... it does immersion breaking silly things like not build a Government Plaza until turn 200.

I often don't build a Government Plaza until turn 200 ...

 

[Leyrann]

I often don't build a Government Plaza until turn 200 ...

Honestly, in some games I still wonder why it's so valuable. Like, is it the governor titles? Out of the buildings only Ancestral Hall (free builder in every new city, for when going wide) and Grand Master's Chapel (only in militaristic games where you also have faith income) even seem good to me, and the loyalty is usually an afterthought if relevant in the first place. Adjacency bonus is nice but not game-changing.

 

[uhu]

Diseases belong in the medieval, renaissance and maybe industrial era.

Right! And because civplayers love to have fun, it should have to be treated in an appropriate way in order to win -- in this case with flagellation, as The Spiffing Brit shows in his entertaining style: Can you beat the BLACK DEATH with MORE DEATH? (Civ6)

Civ6 has an impressive number of entries in Spiffing Brit's collection[]. @Solver: Do you think, we'll see at least 1 cameo of OldWorld there? I mean, he notoriously breaks all kind of games ...

[
Can You Flood The World In Civilization 6 Gathering Storm? Breaking The Perfectly Balanced Game
CIVILIZATION 6 IS A PERFECTLY BALANCED GAME WITH NO EXPLOITS - Unlimited Gold Challenge
Is this the 'Godliest' Way to Beat Civ 6?
CIVILIZATION 6 IS A PERFECTLY BALANCED GAME WITH NO EXPLOITS - Excluding Unlimited Free Settlers
Civilization 6 is A PERFECTLY BALANCED GAME WITH NO EXPLOITS - Multiplayer Production Exploits
CIVILIZATION 6 IS A PERFECTLY BALANCED GAME WITH NO EXPLOITS - UNLIMITED SCIENCE GLITCH is broken!
CIVILIZATION 6 IS A PERFECTLY BALANCED GAME WITH NO EXPLOITS - Excluding Free Giant Death Robots
CIVILIZATION 6 IS A PERFECTLY BALANCED GAME WITH NO EXPLOITS - INFINITE SCIENCE EXPLOIT IS BROKEN!!!
CIV 6 Is A Perfectly Balanced Game With NO EXPLOITS - Infinite Gold, Infinite Science, Infinite Army
CIV 6 Is A Perfectly Balanced game WITH NO EXPLOITS - Gran Colombia Infinite Meteors Is Broken!
NO CITY WIN ONLY CHALLENGE - CIV 6 Is A Perfectly Balanced game WITH NO EXPLOITS Except Maori
CIVILIZATION 6 Is A Perfectly Balanced Game With NO EXPLOITS - Pacifist Only Challenge IS BROKEN
INFINITE PIRATES Is Broken! Civ 6 Sid Meier's Pirates IS A Perfectly Balanced Game With No Exploits
1 TURN SCIENCE VICTORY! Civ 6 Is A Perfectly Balanced Game With No Exploits - Infinite Science!!!
Breaking The Economy With INFINITE MONEY - Civilization 6 Is A Perfectly Balanced Game With Exploits
HOW TO REALLY PLAY CIV 6 - Sid Meier's Civilization is a perfectly balanced game with no exploits
]

 

[Brutus2]

Civ IV was probably the pinnacle for me in this regard although Civ V was pretty good too. Civ VI suffered from too many gameplay elements that the AI couldn't use or used ineffectively.

That's the problem. Civ IV seemed to have good AI because the "stacks of doom" required no real thought. Stack as many of the best units you can muster and throw them against the enemy stack. I don't blame people for wanting a really good AI opponent, but it does bug me when people slam a game because of the AI not being good enough. It's just not a realistic expectation. I have played turn based strategy games all my life and I have never seen any game as deep as CIV with an AI that can think or play anywhere close to as well as a human. If the game is simple enough, the AI can play well but do we want a simplified game with less features so the AI can appear to be good? I love that we now have the 1UPT and "unpacked" cities that might make things harder for the AI but a lot more fun for the player.

I'm sure someone will point out amazing things that are possible in advanced AI but to expect some cutting-edge, super-computer level AI from game developers is just unrealistic to me.

 

[Solver]

Civ6 has an impressive number of entries in Spiffing Brit's collection[]. @Solver: Do you think, we'll see at least 1 cameo of OldWorld there? I mean, he notoriously breaks all kind of games ...

I'm sure he'll manage to do something with OW! We've had some players pull outrageously crazy stuff, which we then fixed but more was eventually found, so I'd be surprised if no game-breaking exploits are discovered in the coming months! Though I don't think anyone's pulling off a victory with 0 cities settled.

 

[Zenstrive]

maybe I am late to this discussion, but the general design of Civ games after VI's latest DLCs seems to give players ample fun while playing it peacefully. So many gamey mechanics are put, especially the secret societies, heroes, evolving barbarians, etc. The AIs somewhat more capable with sea units and now remember to guard settlers, but still will capitulate after 3 of their cities taken and simply not trying to fight the steamrolling.

The trading offers also seem to be giving players more chance to win too.

 

[rocksinmypath]

The thing is, Civ6 already has a framework for this incorporated in Era Score. I'm not saying the Era score system was perfect (like Eurekas, there was a large element of doing non-sensical things just to earn the points in it), but still, the idea that doing certain actions or achieving certain milestones through an era could give you points is a good basis, and victory could be based on this - either through a simple cumulative score, or probably better, through a system where the players with most era score in each era earns points that will decide the victor at the end of the game. Btw. Humankind had some good ideas in this area, even if their system was far from perfect also.

Maybe this belongs in a different thread, but I had a similar vision to this. Sid Meier puts a lot of emphasis on making games feel special as a way to improve replayability of his games, which I think is why it's unlikely Firaxis will abandon victory conditions. I think the Historic Moments feature sets out a good blueprint for how each game of Civ can still feel special without victory conditions. There are many different ways to approach each VC, but if you win a science victory, you just get the usual science victory cut scene at the end, regardless of how you won. I recently played an Aztec game, where I made an early military push against Brazil to enslave all of their units, and after the war, I pivoted to a science victory using the profits from the war. But there's no way for the game to distinguish this approach from if I had just set out to win a science victory from the beginning and turtled my way to it. With a victory point system, there will be a way to track how you earned your points. In the post-game presentation, you can be presented a history book of sort about your civilization, highlighting key moments in your journey.

 

[Leyrann]

Maybe this belongs in a different thread, but I had a similar vision to this. Sid Meier puts a lot of emphasis on making games feel special as a way to improve replayability of his games, which I think is why it's unlikely Firaxis will abandon victory conditions. I think the Historic Moments feature sets out a good blueprint for how each game of Civ can still feel special without victory conditions. There are many different ways to approach each VC, but if you win a science victory, you just get the usual science victory cut scene at the end, regardless of how you won. I recently played an Aztec game, where I made an early military push against Brazil to enslave all of their units, and after the war, I pivoted to a science victory using the profits from the war. But there's no way for the game to distinguish this approach from if I had just set out to win a science victory from the beginning and turtled my way to it. With a victory point system, there will be a way to track how you earned your points. In the post-game presentation, you can be presented a history book of sort about your civilization, highlighting key moments in your journey.

I wonder how the victory presentation would change for my last game, where I had a regular peaceful science victory... except the turn before I won I declared war on every AI I could and nuked them into oblivion.

 

[rgp151]

I've played 4X games for along time. My favorite is still Master of Magic. New modded version of Master of Magic have been put out by fans, including one heavily modded version called Caster of Magic. Besides fixing bugs, the main thing these versions do is seriously improve the AI.

The result of this is that in the original version players like myself were able to regularly defeat the game on Impossible (the hardest level), while under these new modded version, its more difficult to beat the game on Hard (the 2nd from hardest level) than it is to beat the original on Impossible.

When the AI actually makes goo d choices, the recourse advantages are quite difficult to overcome.

The problem with Civ 6 is that you can always win the game by domination once you get the game into the modern era, because the AI simply do anything to make use of its military. Once you can hit cities using Balloons its really all over, and the AI hardly ever make any kind of late game military offensive, so you never need any real defense at home.

But yeah, its been a horrible crutch for so long in 4X games to simply use AI resource bonuses as a ways to increasing challenge. That's never really what anyone has wanted. Even the modded version's MoM don't really do what people want, which is use increasing levels of AI sophistication as the difficultly level increases, like a the way Chess works. e modded versions of MoM are really just all harder across the board, making "Impossible" now truly impossible. But what people want is an AI that does more sophisticated or aggressive things as it increases in level, not simply makes the same dumb choices, but basically with more gold to do it with.

 

[steveg700]

1000x this!
I'm a bit skeptical of victory through Era Score, though. I have a fair number of games where by the late game I'm easily chaining together successive golden ages and loyalty flipping entire neighboring civs almost by accident. Some mods make this a bit better balanced like 6T Grand Eras it's a bit harder to hit a golden age every era. But even then I think this would end up feeling pretty grindy.).

That doesn't speak to a flaw with Era Score as victory condition. That's simply a problem with late-game Civ altogether.

If the AI won't employ planes, they probably can't win a war.
If the AI won't stop clearing all its land to lay down farms, they can't probably can't beat the player to wonders.
If the AI won't employ gurus to keep their debater apostles alive, they probably can't make headway at RV.

And so on. And this impacts era score, which is why the player gets away with chaining golden ages. In the late game, there's no reason the player can't beat the AI to every wonder.

 

[Ashbery76]

I also bring up stats every time this topic comes up. In Civ6, the lower difficulty levels are by far the most popular. .

Is this stat based on players who play it more than a few months? Or based on the 95% that move on to other games after few weeks.I do not believe the core players stay on low setting.

 

[Solver]

It's based on the overall pool, so it does include the players who move on but the stat says the same thing regardless - most people play on low difficulties. Some of them move on to other games, others become hardcore fans but the hardcore fans are themselves a minority.

 

[Leyrann]

It's a bad argument in the first place. The entire point of difficulty levels is to have something for both casual and dedicated players.

 

[Archon_Wing]

Not really. I actually like seeing people break the AI with entertaining tricks. That actually is a skill. Problem is that the AI breaks itself regardless of user input.

I mean, it's historically accurate to have stupid leaders, lol. But they should pose some degree of a threat.

I do expect the AI to be able to carry out a victory if left undisturbed.

 

[TheMeInTeam]

I'm going to preface this by saying I broadly agree (see my earlier post regarding cost of ML vs need for commercial viability above), but will focus on the more-interesting disagreements or building-on points below.

Machine learning in its current shape is irrelevant to Civ games. ML is good at getting AIs to perform a certain task given many - millions and millions - of attempts. In a game like Civ, just coding a few rules into the AI does the same thing. As a silly example, Civ6 has cavalry units and anti-cavalry units. Yes, you could use ML to observe many battles and have it eventually figure out that Horsemen are very poor against Spearmen. But why bother with that when you have designers who already know this and can tell the AI to look at the strength modifiers before it attacks?

ML does have some clear advantages in this regard. Namely, it does not require any developer to be elite or even good at the game in order to create an incredibly effective AI agent.

Yes, a designer knows that spears get a bonus against horses. A designer might not necessarily recognize that in practice, the spear unit class is bad/not worth building regardless even as a countermeasure to horses in some cases, because of how that decision interacts with other mechanics/units. A designer is outright unlikely to understand how all the game's nuanced tradeoffs interact and optimize for them. In most cases, it is impractical/unrealistic for a designer to be as good at playing a game as creating it. They are a) different skill sets and b) require massive time investments, independently, to reach high levels of competency. There are some exceptions out there...IIRC the Celeste dev is pretty darned good at it. But for the most part, a dev won't be doing things like routine 200AD liberalism in Civ 4, and will be worse at looking at a set of potential balance tweaks and predicting the consequences to meta gameplay than an expert player.

ML shatters all that with brute force training. It can, and will, find the best choices to win reliably if given enough training. Better than expert players, better than the best humans in the world. It will even find things humans can't, like how it was at least somewhat variable with two factions in StarCraft 2, but spammed the heck out of roach/ravager timing pushes as Zerg vs top tier ladder competition. I don't think there was a person alive, dev or otherwise, that had solid basis to predict that in advance (those pushes almost always won on the spot too of course).

People want an AI that's fun to play against and that poses a moderate challenge but doesn't win the game. People want an AI that does things to remind the player that it's playing the same game.

Funny things about this quote:

• In the vast majority of strategy games, including every civ game, the preferences in sentence 1 and 2 are incoherent/mutually exclusive.
• My impression is that your quote is nevertheless an accurate assessment of most players.

If an AI were to "play the same game as the player", the AI would try to win...because the vast majority of players, even if they say they emphasize role play or history or whatever else...nevertheless get upset and complain when they lose, and thus attempt to win/take steps to avoid losing, independent of role playing the game state. Going so far as to reject advice on how to play better and want the game to behave differently instead (it's kinda funny how often people blame losses in games like FTL on RNG with some kind of blind spot for how there are multiple > 100 win streaks on a higher difficulty than they play). But at the end of the day, for every 100 win streak player there are hundreds to tens of thousands of "I died to RNG in sector 4" players, and you need people to buy the game to justify making it in most cases...

This bothers me at a fundamental level. It's one aspect of a "typical player" I can't really empathize with. If an AI is not trying to win, then any "challenge" it presents becomes incidental or the consequence of otherwise-inflated bonuses that compensate whatever it's doing in lieu of that. To me, this directly cuts into the experience, and into the meaning of winning myself. Effectively, my opponents aren't trying, and I know it.

I've been around long enough and seen enough different people play/approach civ and other games to know that I'm not only a minority, but a fairly extreme one. But I just can't lie to myself like that.

I don't envy trying to design with this in mind. To try to create coherent game rules/behaviors for players with objectively incoherent preferences.

The entire machine learning idea is a red herring. To make an AI that's more likely to win - or make the player lose - there's plenty of other changes that could be implemented, and I guarantee that most players would hate them. An AI that plays to win will declare war to grab any unprotected Settlers or Workers. It would consistently declare war if you're fighting someone on the opposite side of your empire. It would never trade any strategic resource to you, etc etc. If a majority of players wanted an AI that challenges for victory, it wouldn't be accomplished through ML.

See my first response WRT ML. It's not clear to me, at all, that a highly trained ML AI would *necessarily* do any of these things (though worker/settler steal into killing target is most likely, by my estimation). I strongly suspect that if you make predictions about what it would do, you would sometimes be right and more frequently be wrong.

For example, I doubt a highly trained AI would refuse strat resource swap with distant civ, especially if it (correctly) anticipates that it could utilize the resources it gets in return better than the player. Even in PvP with only one winner, there is some room for diplo, and to the extent it's allowed using it in some capacity is optimal. Top tier players trade in games like Dominions 5, despite that it also benefits an opponent, because making yourself + one other player better off relative to a field of 12 improves your chances to win. I find it difficult to believe that a ML AI with a ridiculous amount of training wouldn't notice that trading is at least sometimes beneficial, and I suspect that absent human emotion that it could do it better than the player.

Same thing about "consistently declaring war" based on unit location. One, it would have to have intel/see that and two, it's not necessarily the case, because this AI might already be at war itself, or in tech deficit etc. Though ML AI would probably not be in a tech deficit.

ML AI would trivially outplay anything a designer can possibly produce. That's not the problem. The problem is that it's overkill and, as you say, wouldn't help designers create an experience that the vast majority of players actually want, which is something challenging but beatable that is playing a different game from them without them noticing it too much.

Not right now, anyway. If they can actually get it to the point where you can have difficulties that keep AI bonuses the same but are instead different ability AI agents, and bring the cost of this down low enough, maybe it actually is worth implementing. That's not a thing right now, to put it mildly, but I wouldn't rule it out in 10-20 years for example.

 

[Leyrann]

I'm going to preface this by saying I broadly agree (see my earlier post regarding cost of ML vs need for commercial viability above), but will focus on the more-interesting disagreements or building-on points below.



ML does have some clear advantages in this regard. Namely, it does not require any developer to be elite or even good at the game in order to create an incredibly effective AI agent.

Yes, a designer knows that spears get a bonus against horses. A designer might not necessarily recognize that in practice, the spear unit class is bad/not worth building regardless even as a countermeasure to horses in some cases, because of how that decision interacts with other mechanics/units. A designer is outright unlikely to understand how all the game's nuanced tradeoffs interact and optimize for them. In most cases, it is impractical/unrealistic for a designer to be as good at playing a game as creating it. They are a) different skill sets and b) require massive time investments, independently, to reach high levels of competency. There are some exceptions out there...IIRC the Celeste dev is pretty darned good at it. But for the most part, a dev won't be doing things like routine 200AD liberalism in Civ 4, and will be worse at looking at a set of potential balance tweaks and predicting the consequences to meta gameplay than an expert player.

ML shatters all that with brute force training. It can, and will, find the best choices to win reliably if given enough training. Better than expert players, better than the best humans in the world. It will even find things humans can't, like how it was at least somewhat variable with two factions in StarCraft 2, but spammed the heck out of roach/ravager timing pushes as Zerg vs top tier ladder competition. I don't think there was a person alive, dev or otherwise, that had solid basis to predict that in advance (those pushes almost always won on the spot too of course).



Funny things about this quote:

• In the vast majority of strategy games, including every civ game, the preferences in sentence 1 and 2 are incoherent/mutually exclusive.
• My impression is that your quote is nevertheless an accurate assessment of most players.

If an AI were to "play the same game as the player", the AI would try to win...because the vast majority of players, even if they say they emphasize role play or history or whatever else...nevertheless get upset and complain when they lose, and thus attempt to win/take steps to avoid losing, independent of role playing the game state. Going so far as to reject advice on how to play better and want the game to behave differently instead (it's kinda funny how often people blame losses in games like FTL on RNG with some kind of blind spot for how there are multiple > 100 win streaks on a higher difficulty than they play). But at the end of the day, for every 100 win streak player there are hundreds to tens of thousands of "I died to RNG in sector 4" players, and you need people to buy the game to justify making it in most cases...

This bothers me at a fundamental level. It's one aspect of a "typical player" I can't really empathize with. If an AI is not trying to win, then any "challenge" it presents becomes incidental or the consequence of otherwise-inflated bonuses that compensate whatever it's doing in lieu of that. To me, this directly cuts into the experience, and into the meaning of winning myself. Effectively, my opponents aren't trying, and I know it.

I've been around long enough and seen enough different people play/approach civ and other games to know that I'm not only a minority, but a fairly extreme one. But I just can't lie to myself like that.

I don't envy trying to design with this in mind. To try to create coherent game rules/behaviors for players with objectively incoherent preferences.



See my first response WRT ML. It's not clear to me, at all, that a highly trained ML AI would *necessarily* do any of these things (though worker/settler steal into killing target is most likely, by my estimation). I strongly suspect that if you make predictions about what it would do, you would sometimes be right and more frequently be wrong.

For example, I doubt a highly trained AI would refuse strat resource swap with distant civ, especially if it (correctly) anticipates that it could utilize the resources it gets in return better than the player. Even in PvP with only one winner, there is some room for diplo, and to the extent it's allowed using it in some capacity is optimal. Top tier players trade in games like Dominions 5, despite that it also benefits an opponent, because making yourself + one other player better off relative to a field of 12 improves your chances to win. I find it difficult to believe that a ML AI with a ridiculous amount of training wouldn't notice that trading is at least sometimes beneficial, and I suspect that absent human emotion that it could do it better than the player.

Same thing about "consistently declaring war" based on unit location. One, it would have to have intel/see that and two, it's not necessarily the case, because this AI might already be at war itself, or in tech deficit etc. Though ML AI would probably not be in a tech deficit.

ML AI would trivially outplay anything a designer can possibly produce. That's not the problem. The problem is that it's overkill and, as you say, wouldn't help designers create an experience that the vast majority of players actually want, which is something challenging but beatable that is playing a different game from them without them noticing it too much.

Not right now, anyway. If they can actually get it to the point where you can have difficulties that keep AI bonuses the same but are instead different ability AI agents, and bring the cost of this down low enough, maybe it actually is worth implementing. That's not a thing right now, to put it mildly, but I wouldn't rule it out in 10-20 years for example.

Do you think then, that it's feasible to train an AI with machine learning for Civilization?

Because I've seen multiple people in this thread argue that it isn't.

 

[TheMeInTeam]

Do you think then, that it's feasible to train an AI with machine learning for Civilization?

Because I've seen multiple people in this thread argue that it isn't.

You would need to bypass the speed limitation. If you tried to just run in the game as-is it wouldn't work.

Other than that, however, are there really 10^26 possible actions for a civ in a game, per turn, for example? That's what a post on page one said was the case for SC2.

It's not clear to me why it's not possible, in principle, to do in Civ 6 if it has already been done in other turn based games, plus games like SC2 and Rocket League. The physics in the latter has to be perfectly aligned into the training games, without actually producing them like in real games. If people are doing that sort of stuff already, what makes it intractable in Civ 6?

I'm not saying it's worth doing for more than curiosity, but I'd like to see a good reason that Civ 6 is special. Unless the engine/code really is some kind of special spaghetti.

 

[vorlon_mi]

Responding to @TheMeInTeam's comment above ... Effectively, my opponents aren't trying, and I know it.

Over in the CivIII forum, a very experienced player set up some AI vs. AI games documented in this thread https://forums.civfanatics.com/threads/civ-ai-vs-ai-gave-it-a-try.674320/ The upshot is yes, indeed, the computer players (a.k.a. the AI) are trying to achieve one of the victory conditions.

In my experience with CivIV (on a lower difficulty than you play), the AI pursue the victory conditions, including Culture Victories and Space Victories. I've not seen them push hard to complete a Domination victory, though they do build large armies and vassalize smaller factions.

In my experience with BERT, the AI regularly build the affinity wonders that would win them the game. They often fail to execute the final victory efficiently, in that I can counter their tactics with better tactics, but each is trying to win the game. I've never seen an AI aggressively take original capitals to achieve Domination victory.

I haven't played enough Civ VI to know whether the AI players are trying to pursue a Space Victory, Cultural Victory, or Religious Victory. Based on what I've read here, they are not pushing to pursue a Domination victory.

To sum up, I believe that the various Civ AI logic does indeed pursue victory conditions that can be *built*, with varying degrees of efficiency. I have not yet seen evidence that the Civ AI logic effectively pursues victory conditions that require sustained, successful military action.