Looks like Civ 6 is done: Kevin called April "final game update"

[Haig]

Limited stackings is the key, no traffic jams but no stacks of doom with 20+ units either.

 

[Pfeffersack]

no 4x has good AI, but some have rules that does not penalize it as much.

if you really want 1UPT, you can give the units more movement points and more open space on the map to compensate, civ6 did the opposite compared to civ5 rules.

and civ4 tactical AI is simpler than civ5, yet more threatening for the humans "thanks" to unlimited stacking.

I concede that "good" is something which has to be defined (and if you equal it to "human-like"...well, then I have to agree - the 4x-Deep-Blue-AI has yet to be created ), however for my taste and in relative terms I can name you 5 games (three of them in modded state though) out of my head which have a nasty good AI:

1) Imperium: Greek Wars ( https://store.steampowered.com/app/1183470/Imperiums_Greek_Wars/ )
2) Pandora First Contact ( https://store.steampowered.com/app/287580/Pandora_First_Contact/ ); I recommend using AIL's (AI developer of this game!) AI mod for the best experience
3) Warhammer 40.000: Gladius - Relics of War ( https://store.steampowered.com/app/489630/Warhammer_40000_Gladius__Relics_of_War/ )
4) Heroes of Might and Magic V (if you use Quantomas AI mod for it)
5) ...and no, that's no typo... Sid Meiers Alpha Centauri. Just use the Will to Power mod (see the civfanatics thread for it here: https://forums.civfanatics.com/threads/smax-the-will-to-power-mod.653694/ ) and the AI will already aggressively expand on the equal-foot-level

Again, a good AI is subjective and please take the list above with the grain of salt them I'm not a too good player in terms of playing aggressively or directed to win at all costs. But I still consider myself having a sense for when AI acts dumb or quite good. In the listed games above it usually falls in the latter class.

 

[Gedemon]

Limited stackings is the key, no traffic jams but no stacks of doom with 20+ units either.

yes, that's how I played civ5.

 

[Archon_Wing]

Eh, Civ 4's AI was better than 5 and 6s in more ways than just combat. It knew how to pursue win conditions and commit to victory, such as switching off research to go culture.

Right now 6's AI struggles to even win the game if there is no war. It can go 100 turns without even trying to wall a city. I'm not even sure if it can win below Emperor. (Well, below Deity tbh and even there it still does a really bad job). It also seems to be completely unable to handle later eras.

I for one, have never expected any AI to be good tactically, but I do expect them to try to play the game in a reasonable manner.

6's AI's failings can be attributable to excessive feature bloat. It was already stupid but adding more systems that it can't handle magnify the issue tons to the point where it just flat out breaks (like in monopoly mode).

Anyhow, 1upt is here to stay, and I feel like we just have to be honest with ourselves that they couldn't make a decent AI either way anyways.

 

[Horizons]

I'm not even sure if it can win below Emperor

Yes, if you don't go to war with it.

 

[Kwami]

I concede that "good" is something which has to be defined (and if you equal it to "human-like"...well, then I have to agree - the 4x-Deep-Blue-AI has yet to be created ), however for my taste and in relative terms I can name you 5 games (three of them in modded state though) out of my head which have a nasty good AI:

1) Imperium: Greek Wars ( https://store.steampowered.com/app/1183470/Imperiums_Greek_Wars/ )
2) Pandora First Contact ( https://store.steampowered.com/app/287580/Pandora_First_Contact/ ); I recommend using AIL's (AI developer of this game!) AI mod for the best experience
3) Warhammer 40.000: Gladius - Relics of War ( https://store.steampowered.com/app/489630/Warhammer_40000_Gladius__Relics_of_War/ )
4) Heroes of Might and Magic V (if you use Quantomas AI mod for it)
5) ...and no, that's no typo... Sid Meiers Alpha Centauri. Just use the Will to Power mod (see the civfanatics thread for it here: https://forums.civfanatics.com/threads/smax-the-will-to-power-mod.653694/ ) and the AI will already aggressively expand on the equal-foot-level

Again, a good AI is subjective and please take the list above with the grain of salt them I'm not a too good player in terms of playing aggressively or directed to win at all costs. But I still consider myself having a sense for when AI acts dumb or quite good. In the listed games above it usually falls in the latter class.

OK, but 3/5 of your games require mods to have "good" AI. And usually, that just means harder AI rather than more fun AI. I think the pattern is clear: 4X games pretty much never have great AI. Sometimes they have artificially hard AI and sometimes they have AI that's fun to play against, even if it never wins, but I've yet to see a game that has both. I guess there are probably a few out there, but I suspect that those games are much simpler, too.

I guess I'd rather a more complex game that gives me more to do, even if the AI can't quite keep up (Civ V, Civ VI), than a simpler game that's hard to win because the AI has bigger numbers (Civ IV).

 

[aieeegrunt]

And anyway, it's not like other 4X games have good AI. They just have lots of bonuses.

Civ4 and Civ Rev are two obvious examples of greatly exceeding Civ6’s AI in just about all areas

Honestly Civ6 might have the worst AI of the entire series

Limited stackings is the key, no traffic jams but no stacks of doom with 20+ units either.

Stack of doom are by far the lesser of two evils

The only way 1UPT isn’t sliding tile puzzle torture just to simply move a unit, is with generous move allowances.

 

[PiR]

OK, but 3/5 of your games require mods to have "good" AI.

So?

It started with this:

I don't know if it's technically impossible to do good 1UPT AI with short AI turn times, or if it's devs incompetence

Which doesn't exclude mods.

I guess I'd rather a more complex game that gives me more to do, even if the AI can't quite keep up (Civ V, Civ VI), than a simpler game that's hard to win because the AI has bigger numbers (Civ IV).

The AI in Civ V and VI has bonuses and still the Civ VI AI is dull when you activate many official modes. I have no such complaint against the AI in Civ IV.
In my case, no, I would rather have a simpler game where the AI provides a challenge.

 

[Pfeffersack]

OK, but 3/5 of your games require mods to have "good" AI.

I was a bit in a rush and not detailed enough with my words. It is absolutely the case with SMAC and HOMM5, but Pandora's AI is even without the mod giving it the final touch already quite decent (as said, the modder equals the AI programmer here - the mod is more of a final AI patch never having released as official patch by the publisher because of...well, probably the typical publisher reason of ending official development at some point )

And usually, that just means harder AI rather than more fun AI.

I have trouble to get what you mean by "hard" (ruthless and designed to win at all cost and/or loaded with "cheats"/boni to be competitive?) vs "fun" (just weak so that everyone can win? Or rather behaving human like in a sense of some variance or caring fro diplomatic relations?) AI.
For the games I listed, I can say that good AI here does not mean "good by many cheats/boni" (not to create another misunderstanding: All those games do offer those "impossible" difficulties, where the AI gets boni left and right. However my impression is solely based on games on the "equal foot levels").
For "being ruthless and having desire to win" - yes, that is greatly increased in all cases. The games vary in their topic and the overall peace/war leaning they have, so in Gladius and HoMM5 there is little "building" anyway. Pandora and especially I:GW maintain a variety in AI behaviour though - if you alone have a look on the modifiers affecting diplomatic relations in I:GW you will be surprised, if you e.g. think that "a lot of things affect diplomatic stance in Civ6" - I think I:GW has more: https://steamcommunity.com/app/1183470/discussions/0/2791621875954477767/ (54, if I counted correctly)...and some of them a pretty "human". Pavel, the dev, said about that system that the intention is to guide the player to treat the AI like a human and not an AI. It's not perfect, but is closer than comparable systems in e.g. the Civ series or Paradox Titles.
Probably the case of SMAC is closest to what you describe - with the mod all AIs expand likey crazy and get strength...and all tend to relentlessy harass you, if you just try to peacefully buider. So yes - here the AI has become more uniform

I think the pattern is clear: 4X games pretty much never have great AI. Sometimes they have artificially hard AI and sometimes they have AI that's fun to play against, even if it never wins, but I've yet to see a game that has both. I guess there are probably a few out there, but I suspect that those games are much simpler, too.

Hard to answer that, as it depends on how you define complex. Also my examples vary in terms of how close the come to Civ. The range goes from SMAC (which is "Civ in space", but closer to Civ2/3 in its mechanics), over Pandora (which has great similarity with SMAC, though lacking some civil aspects like Social Engineering) and Gladius (graphically similar to Pandora, but with a greatly reduced civil part - however, with a deeper combat system), to I:GW (which indeed follows the principle "less features and numbers of different units/buildings/whatever", but what is in is interwoven with everything else). HoMM5 stands hybrid between a roleplay/startegic game a bit aside, as it is almost a stretch to label it as classic 4X.

 

[Archon_Wing]

Yes, if you don't go to war with it.

That's not true either.The AI rarely goes anywhere on Emperor even if I don't do anything.

Example: https://forums.civfanatics.com/threads/is-this-game-winnable.668367/page-2#post-16060598

This is an Immortal game where I was boxed to about 4 cities for the first 130 turns and I didn't attack anyone. I flipped a grand total of 1 city.The AI actually made reasonable moves towards victory and was keeping up on yields and even reached a t4 government but still nothing near with Babylon about to do Moon Landing. I would consider their performance above average here, sadly.

 

[PiR]

@Duke William of Normandy did you see that one?

 

[Duke William of Normandy]

@Duke William of Normandy did you see that one?

View attachment 598678

View attachment 598679

No, because I don't use Twitter. Also, Kevin is really starting to grate my nerves in a good way. He's trolling us, which frustrates but also impresses me.

 

[acluewithout]

Kevin is hilarious. Long may he reign.

 

[Duke William of Normandy]

Kevin is hilarious. Long may he reign.

Long may Kevin the Meme Lord reign!

 

[aieeegrunt]

@Duke William of Normandy did you see that one?

View attachment 598678

View attachment 598679

Wel they certainly nailed the social media angle of making a game.

*draws on pipe*. Well played laddie

 

[Horizons]

That's not true either.The AI rarely goes anywhere on Emperor even if I don't do anything.

Example: https://forums.civfanatics.com/threads/is-this-game-winnable.668367/page-2#post-16060598

This is an Immortal game where I was boxed to about 4 cities for the first 130 turns and I didn't attack anyone. I flipped a grand total of 1 city.The AI actually made reasonable moves towards victory and was keeping up on yields and even reached a t4 government but still nothing near with Babylon about to do Moon Landing. I would consider their performance above average here, sadly.

AI does a fine job of winning either diplo or science in the late game on King. I myself am nowhere near a victory on turn 130.

I'm not a good player though (i.e. not a crazy micro-manager who is close to winning by turn 130)

 

[Horizons]

Just lost a game on King right now. Got a totally hemmed-in start (not even enough space to settle one other city), fought and won early wars, worked my butt off and worked my way up all game only for the AI to turtle away and build a spaceship on another continent with its astronomical production and science cheats.

The AI can win easily, but only in the most obnoxious manner imaginable.

 

[Archon_Wing]

AI does a fine job of winning either diplo or science in the late game on King. I myself am nowhere near a victory on turn 130.

I'm not a good player though (i.e. not a crazy micro-manager who is close to winning by turn 130)

A lot of players aren't winning @ turn 130 normal speed, nor do you need to on any difficulty, so not sure what's your point. And certainly the AI isn't able to complete a victory condition by then.

 

[Jeppetto]

This is so confusing.
It's like the guy is doing some Gollum dual personality scene.

View attachment 598476

There are also Kristina twins on this forum unless they changed their avatars. The avatars were even more situated into position of looking like Kristina just talking to her alter-ego by switchin side on which she stands.

I mean, on a theoretical level I understand the appeal of "challenge," but the AI is considerably less concerning when you're just here to build a nice empire and think of the AI opponents as set dressings...

I mean... that's what Difficulty bar is for. There are players complaining that even on Deity the AI poses not enough of challenge and some players actually want the challenge. Others want to farm. Difficulty bar always presented solution. If I want to build nice empire with AI as side-feature rather than challenge (which still doesn't stop Gilga on friendly mood declaring Surprise War on me) I go for lower difficulty. The only downside I can think off is being ashamed of playing on lower difficulty because of some git gud chadness, which I'd doubt would be prevalent on this forum.

I consider the game on low diff easy enough to be empire builder farm style. If however decent portion of Deity-players don't consider it hard enough, then changes should be made instead of just being "You could just not care". You could just not care too, If the AI will be changed to be actual challenge, it can be diff specific, so it would not affect those who want to farm, while making the game more enjoyable to those who want challenge.

In fact, ideally the difficulties would be split, one using Tactical AI (only react to what it has, easiest AI), one using Planning AI (tactically plans how to get the best few step forth, such as calculating ideal yields of cities, but not making complex decisions, medium AI) and Strategic AI (one that would use some sort of relations to try to wave somewhat logical steps of what way it plans to achieve victory with, adjusting as it goes, it doesn't need to be anything complex, just so that the AI is aware on how to proceed).

 

[Zaarin]

I mean... that's what Difficulty bar is for. There are players complaining that even on Deity the AI poses not enough of challenge and some players actually want the challenge. Others want to farm. Difficulty bar always presented solution. If I want to build nice empire with AI as side-feature rather than challenge (which still doesn't stop Gilga on friendly mood declaring Surprise War on me) I go for lower difficulty. The only downside I can think off is being ashamed of playing on lower difficulty because of some git gud chadness, which I'd doubt would be prevalent on this forum.

Right, which is why I play on Chieftain even though I could play at a higher difficulty if I wanted to.

I consider the game on low diff easy enough to be empire builder farm style. If however decent portion of Deity-players don't consider it hard enough, then changes should be made instead of just being "You could just not care". You could just not care too, If the AI will be changed to be actual challenge, it can be diff specific, so it would not affect those who want to farm, while making the game more enjoyable to those who want challenge.

I think your sarcasm detector could use some work... Although I do think the constant whining about the AI is a tad melodramatic.

In fact, ideally the difficulties would be split, one using Tactical AI (only react to what it has, easiest AI), one using Planning AI (tactically plans how to get the best few step forth, such as calculating ideal yields of cities, but not making complex decisions, medium AI) and Strategic AI (one that would use some sort of relations to try to wave somewhat logical steps of what way it plans to achieve victory with, adjusting as it goes, it doesn't need to be anything complex, just so that the AI is aware on how to proceed).

I agree that splitting the AI seems like an ideal solution.