How does Civ 6 compare to Civ 4?

[Dimplz]

I would love to have the "Always Peace" setting to come back!

 

[Biz_]

civ4 is a much better multiplayer game

civ6 is basically a city-builder instead of a strategy game

all the people who don't want the AI to play to win are in the completely wrong genre. it's a shame that their influence has dumbed down the franchise to this new low

 

[LDiCesare]

I assure you that that is not true. I've been playing King AI and I've just witnessed a war between Gandhi and Gorgo and Gandhi was definitely taking cities by force. It is also very common to see the AI attacking and taking City States. So it can most definitely take other Civ's cities through war. And it takes them even with Walls in place.

I only played Deity after the first run, so maybe with a weaker AI it's better? I doubt it. From what I experienced, and this is confirmed by looking at the code, the I has different city attack behaviors based on whether it's early or late. I never saw a late city takeover, and I've seen it unable to take cities with a lone defender despite overwhelming forces. In Civ IV this would never happen. For this readon, I stopped playing Civ VI and find it impossible to enjoy whereas I love IV. So pretending that the AI in VI (to me: so bad it's not funto play against it) to be on the same level with IV (to me: fun) does not make sense.

The AI should not be designed to game throw. You are saying it should be.

Who says it shouldn't? The AI should be designed to make the game enjoyable because the whole point of the game is to enjoy it. But to me enjoyment comes from immersion, and an AI that suddenly backstabs you because it would lose is immersion-breaking.

If attempting to win makes some victory conditions less viable or not viable at all, that's a condemnation of the design and balance of the game.

Again, no. By your definition, a diplomatic victory is NEVER possible. But some people want it. So some people want things different from what you want, that does not mean they are wrong.

The AI is acting as a player in the game, and as such it should be playing the same game.

No it is not. It is given a personality, a flavor, etc. Some AI's have even been given different rules. They are called city states and they are forbidden from winning from the start.
The AI is not another player. It is an element of immersion, a make-believe. It's here to make you think that you are interacting with a world. It fits the role of an opponent, but it also fits the role of a landscape. It is much more than an opponent.

If chunks of the game become non-viable when AI diplomacy is replaced with humans, there are chunks of the game lacking in design viability.

No. Single player games exist, and Civ is mostly a single player game. Consider any game where the opposition is not symmetric. Does it lack in design? No. It's just a different design choice from what you'd like.

If you make a system someone would actually use in MP, you can then tune AI to use it without it being functionally worthless. That's what we should have seen, but don't.

No. The AI plays more than the role of an opponent.

I've seen attempts to make a case that the AI instead taking on the role of a spoiled child flipping a table is good for immersion, but that case has never been particularly sound.

I can say your case is not sound either. Please remember I said that Civ V went that way and failed. The "I play to win" AI in Civ V did the following: Get killed. Get resurrected by player. After 10 turns, decide that said player is weak because he's at war, so attack him. That would be great if it had worked. Except that civ had 1 city vs. my 10, 5 units vs. my dozens, and was in the middle age when I had tanks. Maybe someone could develop an AI that would be able to pull it, but Firaxis doesn't seem liable to do it ever.

Every civ game has had flaws. The AI's propensity to game throw *by design* was a major one in Civ 4. What Soren said sounds nice, but what happens in practice isn't. 4 is still more challenging than newer entries in the series, simply because of the scale of bonuses and that AI can handle stacking its super discount units in one place better. There is no civ 6 equivalent to 15+ unit rushes at 1000 BC. "Just build 3 archers" is a viable deity survival strat now.

You are wrong again. I got attacked by 4 deity norwegian warriors on turn 12. There was no way to defend agaisnt that. I lost. That's the equivalent of your Civ IV "rush" except it's earlier, so it's an actual rush. 1000 BC is not a time for rush, it's just an early war.

Civ 6 has lots of things in release that surpass its predecessor. The tall vs wide cancer is less prevalent/reasonable, there is a least reasonable depth in builds, and interactivity with situation is something lacking in 5 that is at its highest point now. Those are positive changes.

Unfortunately, civ 6 has not found a way to tackle poor UI, and has only somewhat helped performance. The unit cycling is particularly terrible, and forcing an ini edit to remove that doesn't address the unit control issue entirely. Ranged units got buffed by the terrain change, the exact opposite of what needed to happen with ranged units. Civ 6 is a mixed bag, it also isn't finished. That's not an acceptable thing.

I agree with all this.

Edit: Random events can work, but Civ 4's were a joke and if that's the best we can see in Civ I'd rather never see them again. Hedging against risk with incomplete information is reasonable. Losing 10000 hammers instantly is not, nor is having a random-something that does nothing at all to influence your strategy in any capacity whatsoever.

Yes, Civ IV random events were not particularly interesting.

all the people who don't want the AI to play to win are in the completely wrong genre. it's a shame that their influence has dumbed down the franchise to this new low

It's not the case. The AI designed to win was that of Civ V, and it was much worse than that of Civ IV. The AI should be codedto eb enjoyable, and that means the diplomacy should matter.
It is not the influence of these people which made the AI bad. In fact, it's someone who thought the opposite who introduced game systems he couldn't code an AI for and ruined it.

 

[TheMeInTeam]

Who says it shouldn't? The AI should be designed to make the game enjoyable because the whole point of the game is to enjoy it. But to me enjoyment comes from immersion, and an AI that suddenly backstabs you because it would lose is immersion-breaking.

For this point to be valid, you will have to demonstrate why your immersion is more important than my immersion or anybody else's immersion. If you can't do so, alleging something is "immersion-breaking" is disingenuous. I find the AI trying to play Dungeons and Dragons when everyone else at the table is staring at a Risk board to be immersion breaking.

What is valid? This is a game, a game with defined options, rules, and win conditions. Handicaps for or against it aside, the AI is acting as a player in the framework of that game. Game throwing or table flipping is not reasonable behavior in such an environment and it sure as heck isn't consistently providing immersion for the player base.

You can role play regardless of what the AI does. However, I can't play Civ 6 regardless of what the AI does, if the AI does not play Civ 6.

Again, no. By your definition, a diplomatic victory is NEVER possible. But some people want it. So some people want things different from what you want, that does not mean they are wrong.

Are you trying to tell me that it is *impossible* to design a ruleset whereby diplomatic victory could be a legitimate threat in MP? "NEVER" is a strong word. Civ does have a long history of poorly designed diplomatic victory conditions. That doesn't mean making one that's viable in a competitive environment is impossible for the design team.

So yes, the creation of a victory condition that depends on the other agents in the game throwing it is poor design, it isn't necessary to implement that victory condition in such a fashion, and the AI allowing an obviously poorly designed victory condition because reasons is game throwing.

No it is not. It is given a personality, a flavor, etc. Some AI's have even been given different rules.

People play differently too. Quoted statement does not help your point. Bringing up city states is disingenuous and dodges my point. I won't credit that with a response because it's intellectually rude to the discussion. I obviously wasn't discussing city states, just like I wasn't discussing the AI scripts for automatically exploring, which is also technically AI.

The AI is not another player. It is an element of immersion, a make-believe. It's here to make you think that you are interacting with a world. It fits the role of an opponent, but it also fits the role of a landscape. It is much more than an opponent.

A flower is not a flower, it is an element of immersion...

Actually a flower really is a flower in reality, and what you make of it past that is up to you. According to the rules of Civ 6, the AI (the major civ AI) has the same win conditions as the player.

I can say your case is not sound either. Please remember I said that Civ V went that way and failed. The "I play to win" AI in Civ V did the following: Get killed. Get resurrected by player. After 10 turns, decide that said player is weak because he's at war, so attack him. That would be great if it had worked. Except that civ had 1 city vs. my 10, 5 units vs. my dozens, and was in the middle age when I had tanks.

I don't know what Shafer claimed, but CIv 5 AI is not representative of an AI that attempts to win. For all the allegations that it did, very little in the vanilla game showed any evidence of hard-pushing a victory condition. Even the culture win BTS AI of civ 4, a game that allegedly wasn't trying to win all that much, could put more time pressure on the player than civ 5 AI.

Using civ 5 AI to gauge "play to win" quality is like using civ 5 vanilla release as an example of a game that works well in multiplayer. How much of what an AI did in 5 resembled a halfway-competent human player in the same situation, *even at the most grossly basic strategic layer of decision making*? The answer is "basically not at all". We have yet to see an AI that actually attempts to win the game in an unmodded civ game. 4 and 5 mods both tried to rectify this, and both managed significant improvements...

 

[LDiCesare]

For this point to be valid, you will have to demonstrate why your immersion is more important than my immersion or anybody else's immersion. If you can't do so, alleging something is "immersion-breaking" is disingenuous. I find the AI trying to play Dungeons and Dragons when everyone else at the table is staring at a Risk board to be immersion breaking.

I said "to me". You don't make any such restrictions on your claims, so you're the one being disingenuous. Prove your points first.

What is valid? This is a game, a game with defined options, rules, and win conditions. Handicaps for or against it aside, the AI is acting as a player in the framework of that game. Game throwing or table flipping is not reasonable behavior in such an environment and it sure as heck isn't consistently providing immersion for the player base.

The game is more than game mechanics. Most of the budget in games is about art. Why do games even have music? Everybody can play music at the same time as playing. The game is not defined solely by a set of rules. It's also defined by the ambience it puts the player in. You are forgetting more than half of what the game is in all your statements.

You can role play regardless of what the AI does. However, I can't play Civ 6 regardless of what the AI does, if the AI does not play Civ 6.

No I can't. If the AI behavesin a way that is not believable, like when Civ V decides to attack you when it has no reason to, either in the context of winning or in the context of playing its role, then, no, I can't role play.
Again, you are throwing your assumptions at others. You just don't know what I or others enjoy so please stop making definitive statements.

Are you trying to tell me that it is *impossible* to design a ruleset whereby diplomatic victory could be a legitimate threat in MP? "NEVER" is a strong word. Civ does have a long history of poorly designed diplomatic victory conditions. That doesn't mean making one that's viable in a competitive environment is impossible for the design team.

The idea of a diplomatic victory is to have everyone agree on the winner, without having to win in another way. If the AI plays to win, then you can't ever achieve that because the ai would lose. The whole idea of the diplomatic victory is an "everyone wins" which doesn't make any sense in a competitive game.

So yes, the creation of a victory condition that depends on the other agents in the game throwing it is poor design, it isn't necessary to implement that victory condition in such a fashion, and the AI allowing an obviously poorly designed victory condition because reasons is game throwing.

No it is not poor design, some people like it an ask for it. You just don't like it.

People play differently too. Quoted statement does not help your point. Bringing up city states is disingenuous and dodges my point. I won't credit that with a response because it's intellectually rude to the discussion. I obviously wasn't discussing city states, just like I wasn't discussion the AI scripts for automatically exploring, which is also technically AI.

Actually, to me city states are bad just because they don't play by the same rules as the other players and it's a design deicsion I utterly loathe. But it's still a valid design decision.
You conventiently dodge my point. The AI is not another player. It's a game element.

A flower is not a flower, it is an element of immersion...

Actually a flower really is a flower in reality, and what you make of it past that is up to you. According to the rules of Civ 6, the AI (the major civ AI) has the same win conditions as the player.

And AI are Artificial. They are not players. They are not just opponents. They are elements of the game and most of the time and energy spent on them was not about making them play by the rule but about looking like historical characters, and that included behaving like them. Not making them behave like rulers but like players is not the point of the game.

I don't know what Shafer claimed, but CIv 5 AI is not representative of an AI that attempts to win. For all the allegations that it did, very little in the vanilla game showed any evidence of hard-pushing a victory condition. Even the culture win BTS AI of civ 4, a game that allegedly wasn't trying to win all that much, could put more time pressure on the player than civ 5 AI.

Using civ 5 AI to gauge "play to win" quality is like using civ 5 vanilla release as an example of a game that works well in multiplayer. How much of what an AI did in 5 resembled a halfway-competent human player in the same situation, *even at the most grossly basic strategic layer of decision making*? The answer is "basically not at all". We have yet to see an AI that actually attempts to win the game in an unmodded civ game. 4 and 5 mods both tried to rectify this, and both managed significant improvements...

It does attempt to win. Itjust attempts to win putting some constraints upon itself. In the same way you put artificial constraints on yourself when you play anything but prince, because you lose on production, tech, and whatever other penalty the difficulty level throws at you.
Using Civ V as a reference makes sense. The diplomatic behavior of the AI was the culprit, not its tactical ineptitude. Diplomatic AI is what we're talking about. You seem to say that a Civ game must be designed in a certain way. I think you're wrong, because other people seem to like, enjoy and be willing to pay for a game where the AI behaves like a ruler and not like a player.

 

[Roxlimn]

I only played Deity after the first run, so maybe with a weaker AI it's better? I doubt it. From what I experienced, and this is confirmed by looking at the code, the I has different city attack behaviors based on whether it's early or late. I never saw a late city takeover, and I've seen it unable to take cities with a lone defender despite overwhelming forces. In Civ IV this would never happen. For this readon, I stopped playing Civ VI and find it impossible to enjoy whereas I love IV. So pretending that the AI in VI (to me: so bad it's not funto play against it) to be on the same level with IV (to me: fun) does not make sense.

I don't know what's wrong with the Deity AI. I'm only telling you what I've seen. The AI takes cities. It does so when it's got Varu. It does so when it's got Knights and Crossbowmen. I don't know why the AI in your games has been as bad as that. Sometimes the AI intentionally refrains from taking cities because that's not what it's war is for. I do that, too, sometimes. I don't always need a crappy city. Sometimes I just want to raze things or kill units. Less Warmonger, you know. I gather lots of Deity players are complaining about never-ending Denouncements? Maybe don't take cities? That works if you don't like being denounced.

 

[LDiCesare]

The AI doesn't take city when it has 8 mamluks around it. No mamluk attacks the city. The city and a unit inside strike at one mamluk and kill it. Mamluks move around the city. Rinse and repeat until all ai mamluks are dead.
This same behavior has been reported by others and I've seen the same in several games.
I've also seen a 1-city backward civ declare war on a more advanced, 8 city empire, and not get wiped out despite the other civ having no other war to wage and having lots of units.
The ai does not attack cities most of the time (that is I never saw it do it past classical era).
It makes the game unplayable.

 

[Roxlimn]

I think it sometimes declines to take cities. This appears to be intentional, neither a bug nor an oversight. You can disagree with it. Maybe ask for AI settings that will intentionally take player cities whenever it can.

You're conflating what appears to be intentional design for design oversights.

 

[elitetroops]

My impression is that there must be something wrong with the AI code that causes it to evaluate combat situations wrong and attack way less than it should. This doesn't only apply to attacking cities, but to attacking units as well. In one deity game I took on Kongo with archers. This was a worst case scenario. I had to move my army through thick jungle and a tight choke point, it took me forever to get the units in position to start firing on his 35 strength city, mostly I could not fire from 2 tiles away because of all the jungle, and he had lots of his UU swords that have defensive bonus against archers and can move quickly through jungle to snipe my units. This was one of my first games and I didn't have any feel for how combat plays out yet at that time, but in retrospect I realize that this war should have been suicide. What did the Ngao Mbebas do? Well, nothing really. When my units approached and moved within range for them, they did not attack. I fired a couple of shots at them, then they ran away to heal. Every now and then one would come at me, I fired at it a couple of times and it ran away again. It was just silly.

Barbs are much tougher because they attack a lot.

Intentionally coding an AI to avoid taking cities from another civ they are at war with would be stupidity squared. There is absolutely no situation when an AI should actively avoid taking a city while at war. But clearly there is something in there that causes them to avoid it. The devs themselves said that they have never in any test seen an AI win a domination victory. Maybe this should have prompted them to have a closer look at AI war behavior.

Regarding AI playing to win, I think there's a big difference between AI playing to win and AI playing to prevent others from winning. The AI should play to win. It should know what the victory conditions are and how to pursue them. It should also at some point pick one and actively pursue it. So far I haven't seen any AI come close to a victory, though from what I've read they apparently have some clue about how to pursue a religious victory but don't understand much about the other victory conditions. When it comes to preventing others from winning, I'm with LDiCesare in that there is a limit for how far they should go. If they'd do anything to prevent you from winning, then diplomacy becomes meaningless and all kinds of different AI personalities also become meaningless. Not that either of those would be particularly meaningful in this game anyway... In general I've never viewed the AI in civ games as competitors. They are mainly obstacles. Obstacles that (at least in Civ 4) act in different ways and need to be dealt with in different ways on your path to victory.

 

[rrrwww]

civ 4 with mods is much better than current civ6. However, civ6 has, theoretically, more potential due to just more stuff like civics/eurekas/districts.

Civ 6 also has limitations though which civ 4 did not have: the user interface, very demanding pc specs for late game on big maps, obviously AI but I suspect this will be fixed to some extent, hard to micromanage. There are a few more

All in all, current civ 4 is in a much better state than current civ 6. Current civ 4 with mods like Rhye's Earth, Realism Invictus, and especially fantasy mods all allow for more strategies, possibilities and choices, replay value, balance, and especially AI.

So my verdict is civ4>civ6. And unless the developers seriously change all the limitations of civ6, I suspect civ4 will always be better

 

[CaiusDrewart]

I would love to have the "Always Peace" setting to come back!

The advanced setup menu is one thing Civ IV indisputably did better than Civ VI. Vanilla Civ IV shipped with Always War, Always Peace, One City Challenge, Aggressive AI, Random Personalities, No Barbarians, Raging Barbarians, and a bunch of others. Civ VI shipped with... No Barbarians and No Tribal Villages. That's it!

 

[LDiCesare]

I think it sometimes declines to take cities. This appears to be intentional, neither a bug nor an oversight. You can disagree with it. Maybe ask for AI settings that will intentionally take player cities whenever it can.

You're conflating what appears to be intentional design for design oversights.

I don't think so. If you can make sense of sending units around a city and having them all killed by the player while achieving nothing, then please enlighten me. I'm pretty sure it's unintended behavior.
I believe the ai doesn't want to attack if the unit is going to take, say 50% damage while dealing only 10%. If so, then it should retreat. Or it should accept to suffer some losses, like here losing 5 mamluks out of 8 in order to take the city. Or it should bring archers/artillery/rams/siege towers.
But it makes no sense to have the units remain in range of a city which can destroy them and just move around staying in range till they are all dead.

I accept an AI that is not an optimal player for the sake of character/immersion. But an AI that lets its units die for no reason and without trying to achieve anything I can't understand.

 

[Roxlimn]

I think it's trying to pillage your cities. I've had Russia try to do that, and I let them run around for a bit without killing them and they started pillaging my Districts. It is a sound idea. I do that sometimes just to cripple an AI if I don't want the Warmonger penalties of taking its cities. The AI already retreats far too often when it shouldn't. It's trying to achieve something. It just sucks at it.

I take it you've never tried a pillaging war.

 

[MyOtherName]

In a pillaging war, the AI will pillage, not just loom over your city.

I have had the AI take a city of mine before when I overly skimp on units. The AI actions when it takes a city look just like the AI actions when it's putzing around, except it actually gets around to attacking the city.

I'm pretty sure there's just something else that's overriding/preempting the attack logic -- e.g. maybe archer fire is triggering some unit-preservation logic, and the control flow stops there rather than continuing onto the attack logic. I wouldn't be surprised if this is actually a systematic flaw in the AI; e.g. something overrides the expansion logic, and you find one-city civs in 1400 AD surrounded by vast unclaimed territories.

Alternatively, maybe the "I'm strong enough to take the city" evaluation is extremely over-conservative, and the "I'm strong enough to advance on the city" is not.

 

[Roxlimn]

Pillaging takes movement, usually turns, and the AI can putz around before or actually gets pillaging. If you don't fire, they eventually get around to it. At least that's what I see. Could be an issue with the self-preservation script. It seems a little too scaredy sometimes.

 

[Horizons]

Maybe the AI should be more likely to 'play to win' at the higher difficulties

 

[Emi999]

I'm fine with the AI playing to win, but I disagree that trying to imitate human behaviour is the way to go. I think single player games should offer something more than a skirmish mode (the way rtses do): otherwise they are just dumber humans. You could replace "Ghandi" with "#PLsh1t" at that point and forget about the whole history (i.e. civ) thing.
I guess I got spoiled by SMAC diplomacy system that really worked as I like: flavored non-bipolar leaders able to challenge you if they thought they had the upperhand, but they didn't declare war on you just because "it's time to win". I remember leaders making alliances with the strongest civ on the map so that they could start bullying me. It's not rocket science AND firaxis already did it once.

 

[kornelm1978]

Overall, I think AI agendas were an interesting idea but should be eliminated.

Sid Meier (or Soren Johnson? I forget) said something about this during a game design talk. He said that it's not fair to make the AI react according to rules that the player would not value.

So like let's say you're designing Spain and you make it an AI rule that Spain will never DOW a civ with the same religion. The player can exploit this rule to control the AI's behavior. If he doesn't want to fight Spain, he can be the same religion. Now imagine it's a multiplayer match and your neighbor civ is a human playing Spain. He is just trying to win. He doesn't care what religion you are.

Same deal with Germany. A human German player is not more likely to declare war on you just because you are sovereign of several city states. He just cares if you're in the lead.

The AI should be programmed to just try to win. There are too many AI behavior rules that are in the game purely to show off the "flavor" of each civilization

It was cool in Civ 3, when civs would have different aggressiveness levels or pursued different victory types. That was just making sure you would face a variety of AI strategies instead of every AI pursuing the same beeline. But now after 3 more sequels, the game is cluttered with AI behavior controls that make them behave crazy.

This stuff should be constrained to the city states. Because they are NPCs, they can't win, but they can give quests and favors. It makes sense to give them thematic behavior & motivations.

I agree that CIV that Civ 3 or 4 diplomacy overall was much better and made more sense. I agree with you that, current agendas are too important for AI. But still I believe agendas are not that bad idea in some cases.

While I agree that for Germany - it just spoils the gameplay and it should not be important because it does not lead to win, I dissagre with Spain. From historical point of view religion was important for many ages for all civ and it ads flavour surely. From gameplay perspective there must be "something" in diplomacy. And Spain should/can values religion just a little bit more. If not, if we had just AI as competing "players" there would be nothing like warmongering, friendship, common friends, etc. But all those parameters (like friends becasue of same governemnt, religion, common enemies) over time should be less and less important. The civs should not bother much about common governement in the end of the game while they compete for the space race victory. And for Spain, if it decides to go for religion victory, the agenda works very well with the victory conidtion.

But there are worse agendas - like - "I like becasue of large territory" or "I like you because you have got a lot of gold". What a stupid idea. It does not reflect reality at all and spoils the gameplay. Why in competive game AI values those succesful and more powerful and have good opinion of them while they are threatening their victory, I have no idea. It should be completly the other way round.

 

[MyOtherName]

I'm fine with the AI playing to win, but I disagree that trying to imitate human behaviour is the way to go.

You want the human player to feel like the other civs are active participants in the game: that they are rival players the human fights to surpass.

I don't think anyone is asking for an AI that passes turing test; "imitating human behavior" is mainly meant to contrast with putting the other civs into blatantly non-player roles.

I think single player games should offer something more than a skirmish mode (the way rtses do): otherwise they are just dumber humans. You could replace "Ghandi" with "#PLsh1t" at that point and forget about the whole history (i.e. civ) thing.

The whole history thing is already out the window; in no way does Civ pretend to be anything resembling an accurate simulation of history.

Also, I think you're missing the fact that there is a wide spectrum of human behaviors to emulate -- e.g. lots of people like to play peaceful builder. And even if we aren't trying to make them human-like, AIs often still have lots of parameters to tweak that, when done right, gives them distinct personalities. (e.g. see Civ 4)

 

[TheMeInTeam]

I said "to me". You don't make any such restrictions on your claims, so you're the one being disingenuous. Prove your points first.

All you did was point out "I prefer the AI to game throw", which is why I mentioned it. Immersion as an argument is largely irrelevant and a red herring, because it assumes a source of immersion.

The game is more than game mechanics. Most of the budget in games is about art. Why do games even have music? Everybody can play music at the same time as playing. The game is not defined solely by a set of rules. It's also defined by the ambience it puts the player in. You are forgetting more than half of what the game is in all your statements.

I am not forgetting, I am not mentioning those things because in contrast to the major civ AI, the music and artwork does not take on the role of a player in the game. It doesn't have win conditions. It doesn't engage in actions also available to the player.

On a side note, they could do with putting a little more of said budget into the UI. Even Civ 4's AI remains ahead of Civ 5 and 6 in terms of information, hotkeys, queues, unit controls, and waypoints...and Civ 4's UI isn't good.

No I can't. If the AI behavesin a way that is not believable, like when Civ V decides to attack you when it has no reason to, either in the context of winning or in the context of playing its role, then, no, I can't role play.

If you set the difficulty to 1 I'm sure you can make whatever in-game player choices are to your fancy, and the AI's behavior won't be of much consequence. If you're not doing that, then what you're saying here is not self-consistent. In both cases (AI declaring over flavor or trying to win), you can get declared on at a time you don't want to fight a war. If "role play" means "the AI does what I want it to do and doesn't deviate from that", then role play means you might as well world builder whatever game state obeys your fancy or something along those lines.

When you said "If the AI behaves in a way that is not believable", you were instantly self-inconsistent. America discovering archery in the 1000's BC isn't believable. Hitting late eras before your people know how to do basic things is not believable. The way barb camps behave is not believable. These are game abstractions, and there exists no rational criteria that separates them from your example.

There is one consistent set of criteria for major civ AI: that it plays for a win condition that the game defines. It might not be plausible to real history, but civ 6 isn't as a whole. It would at least be consistent with the context of the game itself.

No it is not poor design, some people like it an ask for it. You just don't like it.

This intellectually rude. It doesn't make sense to quote a point then not address the point made.

The idea of a diplomatic victory is to have everyone agree on the winner, without having to win in another way. If the AI plays to win, then you can't ever achieve that because the ai would lose. The whole idea of the diplomatic victory is an "everyone wins" which doesn't make any sense in a competitive game.

Your definition of diplomatic victory isn't a reasonable conclusion. It does not even fit the previous two entries in the series, where the player could easily win despite one or several opponents not desiring that outcome and voting against it. It has never been an "everyone wins" outcome in civ, it has been "one of the players wins", and there are ways to allow for such an outcome in competitive games.

Actually, to me city states are bad just because they don't play by the same rules as the other players and it's a design deicsion I utterly loathe. But it's still a valid design decision.

City states are not players, per the game's rules. Major civ AIs are, per the game's rules. I'm not the one dodging the issue.

And AI are Artificial. They are not players. They are not just opponents. They are elements of the game and most of the time and energy spent on them was not about making them play by the rule but about looking like historical characters, and that included behaving like them. Not making them behave like rulers but like players is not the point of the game.

The game itself defines their role. You do not define the point of the game. The game's rules and stated win conditions define the point of the game. Simulating flavor or interesting personalities does not necessarily conflict with the game's rules/conditions. Playing a different game entirely than the one presented does conflict that way.

It does attempt to win. Itjust attempts to win putting some constraints upon itself. In the same way you put artificial constraints on yourself when you play anything but prince, because you lose on production, tech, and whatever other penalty the difficulty level throws at you.

This analogy is non-sequitur and nonsensical. The motivation, decision process, and context are all different.

Using Civ V as a reference makes sense. The diplomatic behavior of the AI was the culprit, not its tactical ineptitude. Diplomatic AI is what we're talking about.

The reason I said using Civ 5 doesn't make sense is because the diplomatic behavior of the AI in Civ 5 better resembles abject game-throwing than it does something even a mediocre player might choose when trying to win, let alone a good one.

You seem to say that a Civ game must be designed in a certain way. I think you're wrong, because other people seem to like, enjoy and be willing to pay for a game where the AI behaves like a ruler and not like a player.

People seem to like, enjoy, and be willing to pay for a game where the UI requires more inputs is necessary, the game often has to re-sync in multiplayer, and the unit cycling that is on by default is broken outright. People bought previous entries in the series where the game's UI routinely lied to the player, and despite that the game has hidden rules (fake difficulty). That doesn't mean any of those things make the game better, or raise its market value one bit.

People buy civ for different reasons, from historical theme to strategic choices presented to a limited tactical combat system. None of these things, not even the AI having a historical flair, require the AI to routinely game throw, just as they don't require the game to hide its rules.

You're conflating what appears to be intentional design for design oversights.

That line gets blurry sometimes, especially if the developers intentionally neglect to provide information about the game's rules. Sometimes what is design intention vs not-intended outcome is unclear.

The devs themselves said that they have never in any test seen an AI win a domination victory. Maybe this should have prompted them to have a closer look at AI war behavior.

The problem is that any change in AI war behavior instantly sets at least some of its opponents to also make similar choices. Given the sheer number of units involved you would expect this condition to be rare for the AI vs AI games, unless you disable the others.

When it comes to preventing others from winning, I'm with LDiCesare in that there is a limit for how far they should go.

The important point is where you set that limit. An obvious self-evident minimum is that it should not lower its own victory odds in an effort to hamper the player (an annoyance that some "flavor" AIs actually do). After that it becomes less obvious. Given the game's rules, it should generally try to resist religious conversions, sabotage space progress, and all things being equal bias attacks towards nations that are close to winning.

The fact of the matter is that Civ 4, 5, 6 diplomacy is permanently incomplete/incompatible with the game until they make it compatible. Joint victory was possible with permanent alliances in civ 4, but in most civs the game purports a diplomatic system but then undercuts most of its potential by making the win conditions favor abusing it or limiting interaction with it. That is *not* a design necessity.

Maybe the AI should be more likely to 'play to win' at the higher difficulties

Aside from the generic "make it better overall", it's problematic to write different AIs at different levels then implement new mechanics/civs/rules later and wind up writing multiple scripts. It makes a lot more sense to write your best AI right away, then tune the bonuses in difficulties to how it performs.

The AI WILL usually try to off you fast on deity right now, not that it's particularly good at it.