Whatever they do I hope the AI is substantially upgraded

[Kaan Boztepe]

I believe we are bashing the AI a little bit too hard here. I think it works fine for the religious victory as well as the culture victory. domination and science game is weak , i agree but it is good at building stuff ( exception being the ridicolous wonders in cities that dont benefit at all ) . Also they ally themselves against warmongers although since they are weak at army control ( domination game ) this aspect doesnt benefit them. overall the bad domination game/army control hurts a lot since there is no penalty for going wide and a lot of incentives the AI falls behind the longer the game continues. Hopefully if they or the modders can get that part just a little bit better the game would become a lot more challenging.
To repeat, what a lot of people are saying, is the community here in these forums are part of the 5% that can win at higher than prince difficult might cause all this bashing.

 

[Disgustipated]

Challenging games have been good because they are absorbing, intriguing, and respectful etc. In the late 80s one of the things games could offer was challenge. There was no great technology, graphics, animations. Games were simpler, but still that one feature could absorb people for so many hours.

I found games too challenging back in those days, though some of the games I was still pretty young. I never finished Bard's Tale 2 and 3, or Wasteland. Had no internet back in those days, so if I couldn't figure out what I was doing wrong, I was never going to finish. Even Fallout I got stuck in one area of the Glow. Although that was more of a graphical issue because I couldn't see an opening because it was so dark, wasn't really a problem with my ability. Luckily by then the internet was around and I could look that up.

I don't want a massive challenge, I just want an AI that does sensible things, not boneheaded things that not even a child would do. It's okay if they don't play optimally, I don't, so I'm glad the AI doesn't. I just don't want to see an AI do things like build Petra in their only desert tile.

 

[darko82]

I don't want a massive challenge, I just want an AI that does sensible things, not boneheaded things that not even a child would do. It's okay if they don't play optimally, I don't, so I'm glad the AI doesn't. I just don't want to see an AI do things like build Petra in their only desert tile.

If you don't want a challenge, do not choose diety. People who choose emperor or beyond probably should expect some reasonable challenge.

Not challenging or replayable games get boring too fast.

 

[Aristos]

I think people still want a challenge.

I don't. Not most, anyways. Simple argument: if that were true, civ 6 wouldn't have sold as it did.

I don't need to plan strategy or use available options if things aren't working as supposed to. I can blindly hit the next turn button.

And you think they did not do that on purpose? Because I don't know, and as I said somewhere else, I cannot disregard the simple possibility of incompetence (and when I said that, a few that usually, coincidentally or not, defend the developers no matter what, came after me like rabid dogs... wait and see them coming). Especially after seeing them play, and commenting on their own play, in their live streams. But I also cannot disregard the possibility that it is on purpose, knowing the majority of their new audience.

 

[darko82]

I don't. Not most, anyways. Simple argument: if that were true, civ 6 wouldn't have sold as it did.



And you think they did not do that on purpose? Because I don't know, and as I said somewhere else, I cannot disregard the simple possibility of incompetence (and when I said that, a few that usually, coincidentally or not, defend the developers no matter what, came after me like rabid dogs... wait and see them coming). Especially after seeing them play, and commenting on their own play, in their live streams. But I also cannot disregard the possibility that it is on purpose, knowing the majority of their new audience.

New audience would explain the new cartoonish artstyle? Dissapointing. Even though, I did not like the Civ V graphics. I prefer Civ VI ....

 

[Aristos]

New audience would explain the new cartoonish artstyle? Dissapointing. Even though, I did not like the Civ V graphics. I prefer Civ VI ....

New audience explains everything. What @TheMeInTeam likes to call "doormat market". I prefer to see it as the instant gratification generations market (which also explains everything, and not only in gaming)...

On the other hand, plain incompetence would also explain a lot, that's why I cannot disregard either... hard to say, really.

 

[kaspergm]

I still think the question is a bit more nuanced than put forth by some here. Yes, I agree most players these days - myself included - wouldn't like it if AI was too tough. You can call that instant gratification. On the other hand, if you programmed the AI to strictly go for victory by all means, the game would deteriorate into something I don't think is true to the core of the game. If AI should try to win the best it could, it should always spam warriors and rush you instantly. In the current frame of the game, if AI always started spamming wonders and rushed you immediately, it would probably win 95 % of games because of the headstart it gets (on higher levels at least). Would that be fun? Not really. Arguably then you could take away all AI bonuses to make starts more even, any maybe at some point you'd reach a point where player could withstand initial rush, but then what would happen - AI snowballing would start even earlier? I'm not sure.

Likewise in late game, optimal strategy for AI would obviously be an all-out war to nuke everyone. Pretty much like some players play the game. I'm not saying it's wrong, it's just not the game I like to play.

That's not saying that I think AI is great as it is, because I don't, on the contrary, I think it's rather abysmal. The fact that only barbarians know how to use their units is disheartening, and the AI seems incapable of both an effective science strategy (it will research everything and then fail to recruit GP to help build spaceship parts, it will even or used to at least, fail to repair sabotaged spaceports), an effective culture strategy (not sure if it even knows how to theme great works or artifacts) and an effective religious strategy (it will send in single apostles or even unguarded missionaries, which actively works against itself). But how to define the line between an functional AI and one that still seems somewhat sensible in terms of diplomacy is not easy for me to see.

 

[Gorbles]

@darko82

Imagine thinking stylised graphics are down to "the kids" (which seems to be the default stand-in for "new audience" when people mention it) instead of a growing realisation amongst industry professionals (and marketing, and so on) that the endpoint of hyperrealisation in graphics is a) unsustainable and b) not actually desireable wrt. psychology and the uncanny valley effect. Nevermind scaling that to a 4x or other TBS game where you have to fit so much on the screen at once.

I wonder if you think the same about Endless Legend? Heavily stylised, definitely complex, definitely the same kind of market.

We seem to be in an odd age. We live in a world where we demand more complicated mechanics, more polished mechanics, higher visual fidelity, but when the one thing that isn't any of those (the AI) doesn't perform up to scratch (because it's rarely one of the things asked for) we get these weird sweeping generalisations about who games are made for, in some amazingly negative context.

 

[Kaan Boztepe]

the fact that the barbarians actually fight not only passable but good is a good indicator that the AI for the civs is either taking other things into consideration (like defending territory or cities/units ) or is handicapped. So we can assume it is a choice done by whoever developper/publisher.

 

[Gorbles]

Barbarians have less to calculate, and less things blocking their actions. They have no alliances, no diplomatic courtesy, no Cities of note, no infrastructure to defend. They can literally just always take the offensive action. They appear to defend their camps to some extent, but it doesn't appear to be an overriding thing.

 

[Disgustipated]

On the other hand, if you programmed the AI to strictly go for victory by all means, the game would deteriorate into something I don't think is true to the core of the game.

Actually what may be interesting is have 2 modes of AI. Competitive and Role playing. Because sometimes I'm in the mood for both, would be neat to have both types of AI. The first would be an AI that is programmed to win at all costs. Ignoring things like even their agenda and districts that don't help with their favored win condition (each civ should have this designated). The 2nd would be the role playing AI, similar to how it is now, but with things like agendas fully fleshed out to create a personality, not just a hard 0 or 1 for satisfying an agenda.

 

[ezzlar]

Given that the AI will always lag behind maybe some rubberbanding would both improve immersion and difficulty. The only rubberbanding feature we have today is harder to reach golden ages the more you have of them, and that is not so punishing. The rest is mostly snowballing rewards.

I liked that Rhyes in Civ 4 actually punished overexpansion (stability and science penalty), exponential maintenance costs, economic stagnation penalty etc made you really think about how far you should push your empire. But many seem to prefer that successful empires should only lead to more successes.

 

[SupremacyKing2]

I liked that Rhyes in Civ 4 actually punished overexpansion (stability and science penalty), exponential maintenance costs, economic stagnation penalty etc made you really think about how far you should push your empire. But many seem to prefer that successful empires should only lead to more successes.

I think the key is finding that balance between both. You do want expansion to be rewarding and you do want success to lead to more success but at the same time, the game still needs to give the player a challenge. If you get so big and powerful that there is no difficulty or challenge anymore, that's not desirable. The risk with snowballing is that the player won't have any challenges anymore. Now if the game had another civ also snowball on another continent where they could be your rival in the late game, like the US and the USSR during the Cold War, that would be cool, but it rarely happens in civ. In many games, when you start to snowball, you just snowball even faster and there is no challenge at all. That's why there needs to be some anti-expansion mechanic, it just needs to be balanced so that it does not kill expansion altogether. I feel like civ5's global happiness was too harsh against expansion but civ6's amenities are not quite hash enough. But overall, I do prefer civ6's more pro-expansion design philosophy, it just needs to be tuned a bit so that it is not quite as easy.

 

[HorseshoeHermit]

Contents of this post:

1. Response to @Disgustipated
2. Agendas: Maybe not a mess
3. Snowballing

---

Actually what may be interesting is have 2 modes of AI. Competitive and Role playing. Because sometimes I'm in the mood for both, would be neat to have both types of AI. The first would be an AI that is programmed to win at all costs. Ignoring things like even their agenda and districts that don't help with their favored win condition (each civ should have this designated). The 2nd would be the role playing AI, similar to how it is now, but with things like agendas fully fleshed out to create a personality, not just a hard 0 or 1 for satisfying an agenda.

Not an insane proposal, considering that "Ruthless" setting has been a thing in the past.

In my view there are five true difficulty settings:

Training wheels
Casual
Normal
Ruthless
Deity

Ruthless and Deity aren't on the same scale, exactly. Deity still "handicaps" the AI with its historical behaviours, its personality. You're playing the regular game but among giants - and yeah, Sid Meier's Civilization is primarily a single-player, historical themed RPG experience. The sales numbers have made it so. Normal and Casual could be Casual and Hard, the important thing is just to give a basic setting difference for those who "want" to lose and those who'd rather not fail. And Ruthless is the single setting where the AI plays to win. No point putting knobs and sliders for this behaviour - the goal is clear. Try to simulate a smart opponent.

The handicaps for Ruthless would be as Meier figures, from that panel talk he gave on AI. Like, players actually expect an AI to cheat a little even at the 'fair' setting, "because the Human can reload a game to get what they want". I'm not saying I agree with it, I'm just capitulating to some unchangeable facts.


---
I think the Agenda system is a good one from the perspective of iterative design, and its problems are not too far from solutions. Like, one huge one (which I think someone posted up about elsewhere) is that an AI should not judge any lack of yours until their own performance in that metric is better. So Trajan can't hate you for having a small empire until he makes his larger. Of course this only is for the 'sympathetic' agendas and not the 'contrary' ones like Emperor Qin or Pedro. This fix is well within the infrastructure's ability to measure - just measure the AI too. A matter of writing the logic, really.


---
I want to challenge the idea that strategy game always means success must breed success. I dunno. It looks like an inevitability, even the definition of a strategy game, but I have become more and more desperate to find an exception. A game that intrigues me is one called Through the Ages designed by Vlaada Chvatil. In that game, somehow, the progress of ages still keeps players within reach of each other, yet also your choices definitely can be failures. It must be partly due to how "tech" progress (mechanically, just "progress", of a card-rotisserie backbone) comes down to all the players equally (though their ability to employ what they draft is delimited by earlier success). A worthy study, in my opinion. I despise the idea of a World-tech tree in Civ every time I see it, but the structure of the proposal, of having one, pillar backbone for all player's growth, that must, must surely, be the key.

Another place to look is to RTSs, where the data is clear. 4X can be diverted from snowballing by balancing around one, constant resource: attention. Starcraft is commanded through clicks, and clicks can be enhanced with skill, but an upper limit is reached in Human possibility instead of having ALL resources act like textbook economic capital, multiplying without fail. Of course, a TBS has no concept of attention, but the search becomes for something that has the same quality, which can be used as a pillar for that balance in the same way.

Alternately, a game could simply be designed to END FASTER, in response to the snowball. It's only a history of the turn-grinds in this franchise that make us look for the problem elsewhere, as if we can't change that. If your win is inevitable, the system must be made smart enough to tell that you've won... and let you win. Of course, a nearby, preferable option to that would be using some kind of dampener or downscaler to reduce the effects of the snowball to within the same magnitude as the effects of ongoing skill and stewardship. Perhaps looking to the RISE AND FALL of empires as a thematic inspiration for such effects. If only Civ could attempt such a thing. . .

 

[SupremacyKing2]

My current game has been rather enlightening when it comes to the AI. Mapuche has conquered city-states like crazy on my continent. In one case, he was rather persistent. He conquered Jerusalem which triggered an emergency. I liberated Jerusalem. A few turns later, he conquered Jerusalem again. This time, I declare war again and capture Jerusalem and keep it. Then, now in the modern age, Mapuche sends a decent military of artillery armies, armor armies and a helicopter army by sea around my empire and makes an amphibious landing and captures Geneva! Unfortunately for the AI, Geneva was deep inside my empire so it flipped almost instantly to me. So the AI basically gave me a free city. The moral of the lesson is that the AI is clearly competent at capturing cities since it can send a sizeable force to capture a city and even make an amphibious landing to do so. It also knows how to form corps and armies since almost its entire military are army units. But it failed to see the "big picture" that capturing that city state so deep in my territory would flip and as a result, give me a free city. The AI is simply following its logic to capture city states at all cost. So I would argue that the failures of the AI is not so much that it can't perform the necessary functions like attack a city or form corps/armies (it clearly can) but that it is following a particular logic (like capture city states) without considering the higher level strategy.

Another interesting scenario a bit later in my current game. Mapuche declares war on me (I am occupying a bunch of his cities). He swarms me with a sizeable military of armor armies, helicopters, artillery armies and spec ops. Now, I have 2 cities and two military encampments pretty close together so I am able to do a bunch of ranged attacks. But my mech infantry units are not able to stand up to his armor armies so I am just using my ranged attacks from my cities, military encampments and artillery armies to try to hold him off. But at one point, the AI brings a bunch of units including 3 artillery armies up to my city on the far side of my encampments where they can't range attack. I was sure that the AI would take the city. But instead the AI moves his entire military passed my city and tries to attack Jerusalem which I own. But his military is now in range of both encampments and both cities. My ranged attacks really soften him up and he retreats.

So again, the AI showed that it is capable of amassing a strong military. It knows how to form army units. It has a diverse military with both melee and ranged units necessary to capture a city. The AI had armor armies, artillery armies (3 of them!), helicopters and spec ops. So why didn't it attack that first city with everything it got and take it? I am not sure but I think the AI could have taken that city if it had done a coordinated attack with all 3 artillery armies at once and hit the city with its armor armies too. But instead it moved past my city to attack another city and also attack some of my isolated units. Maybe it was still following its logic to attack city states (since Jerusalem used to be a city state) or maybe it knew that it actually could not have taken that city? Maybe it was following a different logic to destroy my military to prevent a counter attack? Or as others have stated, perhaps the devs have deliberately programmed the AI not to ruthlessly capture your cities because they want the player to win and have fun and so they don't want the AI to crush the human?

In conclusion: the AI is actually competent at the basic functions like forming corps/armies, amassing a large military and even capturing cities when it wants to. I would argue that the AI's incompetence stems from 2 things: 1) it follows a role-playing logic (like capture city states) without considering the bigger strategic picture and 2) the AI is deliberating programmed not to crush the human player when it can.

If I am right then the reason the AI is "weak" has more to do with the devs' design philosophy rather than lack of effort into the AI. In other words, it's not that the devs are incapable of making a good AI, they just choose to make an AI that is good in terms of their design goals for the game but that unfortunately, is also easily taken advantage of by competent human players.

 

[ezzlar]

One thing I dont like about AI behaviour is the combination of agendas and that they are playing to win. It should be one or the other. Or some kind of balance. They all appear crazy now.

 

[Victoria]

the AI is deliberately programmed not to crush the human player when it can.

As has been shown, even on deity when the AI can win it waits until T300, I have many times had near death cities the AI has walked away from.
I can walk troops between 2 enemy city walls and only 1 will shoot at me pretty consistently, tell me that is not either intentional or a timeout.

This thread pops up with such consistency and the arguments are always the same. I see some pretty solid proof that the AI has been messed with and so the requirement for a smart AI is moot.
If this game was to review all possible moves it would time-out
The design has no memory, the AI cannot remember your scout popping into view them backing off last turn. Try to be smart with no memory.
Real war was often not smart anyway, charging cannons with cavalry consistently failed and yet is repeated in many famous battles where we find the reasons for the actions are miscommunication, pride and stupidity.

The AI is better than it was, no doubt. If you want a proper challenge play multiplayer but expect gamey play because that is smart play.

Currently, you can't do a THING about it.

Of course you can, a touch exaggerated in my view. If you are not close and have no suze then you cannot. That makes perfect sense.
If you are close you can protect it
In my last game I could see how important Stockholm was going to be for me and how vulnerable it was so I suzed it then upgraded its troops and 10 turns later it was attacked which I easily fended off. At 30 turns they had walls and a good army, this often stops the AI however I checked them each turn to make sure they were OK.
As others will have mentioned there are mechanics in the game that work only makes if some CS have been taken over so there is intent, it is up to you to protect these most valuable resources rather than sit back, do nothing then claim in caps in a thread that nothing can be done.
At worse, attack the taken over CS and free it for 6 envoys (3 if early) and also a good rep with other civs.

 

[HorseshoeHermit]

Chess games used to be popular until they thrashed the average player on simple levels.

You've completely ignored the context of that one sentence, on top of erasing it from your quotation, in such a way that I find this neglect to be vicious (objectively) and a personal slight.

 

[Victoria]

You've completely ignored the context of that one sentence, on top of erasing it from your quotation, in such a way that I find this neglect to be vicious (objectively) and a personal slight.

I apologise, not intentional, I’ll remove, I liked your post, did not agree with all of it but do not want to cause offence.
“The world is yours for the taking” is something the AI does to tell you you have won before time.

 

[Infixo]

@SupremacyKing2 The behavior you describe in the post is a result of how Operations and partially Tactics work in Civ6. It has nothing to do with "role playing" (there is no such thing in Civ6) nor "not to crush human player" (I wish it was that simple).

Every city attack the AI performs is a separate operation that recruits its own units. The target for the attack is chosen based on city valuation and distance. Issues:
1. No bigger picture when running a war - this one is true. Each operation is independent and runs as long AI calculates it can win. Once it determines it cannot win, it either retreats or looks for another target.
2. The vanilla game allows for operations as far as 30 tiles away from you. That explains why AI makes stupid attacks deep into your territory. It didn't matter in vanilla because you could keep that city buy with R&F it changes drastically as such a city flips over anyway.
3. City valuation is quite complex, but in general Original Capitals are valued 200-300 pts. more to incetivize Dom victory. It so happens that each Minor is also an Original Capital and that is why AI values them so much (at least early and mid game, later this value is less important). There is no easy fix for that as there is no parameter to distinguish Majors from Minors when running an operation.
4. The actual behavior on the map is controlled by Behavior Trees. In one of latest patches Firaxis added a shortcut to faster take cities called "CanAlphaCity". I suspect that there is a bug in the implementation here that AI only executes it correctly when a melee unit is adjacent to the city. It is how the attack was performed before that change - melee units are surrounding a city, AI takes its health to 0 and then captures the city.