Likelihood we'll get a decent AI

[Dale]

Old World was developed with the intent of being a game the AI could play. Per the discussions above, based on past experiences with the current Civ dev team, it is unlikely that this will be an important consideration for them. If they come up with a mechanic that they think players will find fun, they're unlikely to exclude it from Civ 7 just because they can't teach the AI to use it properly.

I can tell you with a high degree of certainty, that your statement is just plain incorrect. Old World design has never been limited by the AI, each new feature was added, and the AI made to understand and use every new feature.

 

[Trav'ling Canuck]

I can tell you with a high degree of certainty, that your statement is just plain incorrect. Old World design has never been limited by the AI, each new feature was added, and the AI made to understand and use every new feature.

Okay. I was going by what others have told me about the Old World game philosophy (design new features with the AI in mind). I have no personal knowledge as I have not been part of their dev team or playtest group.

 

[InsidiousMage]

Wait, where is this coming from?

A bunch of places. As already mentioned, Steam achievements for difficulty setting skew heavily towards easier difficulties. But its also a general trend in the industry. The dev team for Hearts of Iron 4 have started that the majority of players use lower difficulty settings, for example. The simple fact is that the vast, vast majority of video game players, console and PC, are "casual" in the sense that they play games to have fun, however they define that. People who play for a challenge or to constantly improve at a game are a small minority but they are also the most vocal and tend to show up more frequently on forums and so on.

 

[Dale]

Okay. I was going by what others have told me about the Old World game philosophy (design new features with the AI in mind). I have no personal knowledge as I have not been part of their dev team or playtest group.

When new features are discussed, the AI is always considered how it will impact the AI, and how the AI can be made to use it. But the AI has never limited or dictated the features going into Old World. First and foremost, the game needs to be fun for players. Then the AI is considered how it is going to use it.

But I do agree with your comments about Civ6. It felt (to me) like features were added because it was considered fun for the human, but then never worried about how the AI was going to be made to use it.

 

[Arcaian]

Wait, where is this coming from?

Even with people who use save games for achievements, or just unlock them all, or set up extremely biased starts for easy achievements, you can see that it's pretty rare to play at higher difficulties from the steam achievements. About 75% of players have played more than an era of the game (75% of players have built 6 improvements or changed government). 37% of players, or about half of all players who have completed an era, have won the game on any difficulty. 30% of players have won on Prince - about 80% of players who have won the game have won it on at least Prince difficulty. Just going to king, that number goes to 15% - half as many people have won at King as at Prince, and we're now down to about 20% of the people who have played the game for an era having won at King difficulty. From there, it's at 10% for Emperor, 7.5% for immoral, and 6.4% for deity. The biggest jump is that prince -> king change, then king -> emperor, and if you get all the way to emperor, you probably got on to win at deity as well (again, assuming people aren't using less-legitimate ways to get achievements). I think it's fair to put 3 broad categories of players from this:

• The majority of players, who play at about Prince difficulty - it seems lower difficulties either don't get used as often, or more likely in my eyes, players start at lower difficulties and work up to Prince, then stop. It's likely perceived as the 'fair' difficulty too - you get no advantages, the AI gets no obvious advantages (they actually get boosted XP, a free tech boost, and yields, but that's much less visible than them having a bonus on all the combat strength calculations, and I guess extra warrior at turn 1 if you spot them there).
• A group of people who push a little further, but aren't trying to get to the super high difficulties. King is definitely included in this - the majority of players who complete a game on King difficulty don't go on to Deity, and about half go to Immortal. I don't think I'd include Emperor here - the drop-offs get much smaller after emperor, so this is just for the King players.
• The people pushing to complete the game on the hardest difficulties, who have completed Emperor+ difficulties. Most people who complete the game on Emperor will also complete it on deity. I imagine this is where most people on CivFanatics fall into - not all of us, but a good deal more than the 10% of the population that achievements say this makes up in the general user base. This is a surprisingly large group if you trust these achievement stats - about 25% of people who have completed a game have won a game at Emperor difficulty.

And I guess the fourth category of "didn't finish a game for whatever reason", but they're not super relevant to this analysis. The big confounding variable here is how many people are using save games/the achievement unlocker/biased starts for achievements. We can estimate the second - for people who have unlocked all achievements, they can't be more than 0.2% of the achievement base, as that's the least-performed achievement. Even if we say that people did this unlocking before the DLC achievements became available, they can't be more than 0.9% of the achievement base, as that's the rate at which the most uncommon Vanilla achievement I saw is used. Save games and biased starts are extremely difficult to estimate - I know some people want to say they've beaten Deity and so will do a high sea-level archipelagos map with a naval power civ on a duel map against someone who isn't very militaristic. I don't think the inclusion of data on this, if it existed, could change the analysis in favour of there being more high-difficulty players; the ratios get worse for the proportion of the game population that players at Emperor+ difficulties as more people are included this way, so the numbers above represent a best-case situation for a large proportion of the game's population being interested in the high difficulties.

 

[Keo]

I also hope the AI is improved greatly and think it’s the most important thing to be addressed. However my hopes are much simpler. I’d still be happy if they only improved the military side of things (although anything and everything can always be improved). I just want effective navy’s and air forces. Armies that use all their units to their strengths and know when and how to attack or retreat. For example if an army’s infantry and cavalry get knocked out, the siege weapons should not be still advancing alone to their deaths as it currently plays out.

 

[Leyrann]

• A group of people who push a little further, but aren't trying to get to the super high difficulties. King is definitely included in this - the majority of players who complete a game on King difficulty don't go on to Deity, and about half go to Immortal. I don't think I'd include Emperor here - the drop-offs get much smaller after emperor, so this is just for the King players.

There's also a simple gameplay issue that may contribute to the drop-off between King and Emperor. Emperor is the first difficulty level where the AI receives a second Settler at the start of the game, which makes the early game significantly more difficult for the player, and may also lead to players abandoning games because they notice that they're behind.

I remember in the early days, I found the jump from King to Emperor quite difficult myself. (nowadays even Deity just doesn't phase me, but like, I'm in the civfanatics category obviously)

 

[Siddorm]

I generally play on higher difficulties but I rarely complete a game because I prefer starting a new game over playing through the late game.
If the AI were better at higher levels (and not just awful but with larger bonuses) and if the late game were more engaging, we would perhaps see a higher proportion of players winning at higher difficulties.
The people I know in real life that play this game (and they are not "fanatics"), all say the AI is awful. The only time I have ever heard anyone say otherwise has been on these forums.

 

[Solver]

Wait, where is this coming from?

Aside from the other sensible factors, this is clear from numbers. CFC has ~340k accounts, which includes every fake/spam account, people who have never posted, people who haven't played Civ for twenty years, etc. 2K doesn't release sales numbers other than some broad aggregates every few years, but Civ6 sales are in the ballpark of 10 million copies.

If every CFC account was a Civ6 player, that would still only be about 3% of the customer base. The site's name really is accurate - if you've made a single post here, you've done something that only 1-2% of players do.

 

[Quintillus]

With the "Old World test" as the thread-established benchmark, I think it's pretty unlikely we'll get a decent AI. Old World's AI can make me sweat and have to scramble in a way that the Civ V/VI AI never has.

But Firaxis is taking their sweet time with Civ7, and maybe part of the reason is that they finally realized that a lot of people find recent Civ versions' AI to be boring after the first few games. The majority? Perhaps not. But not a negligible number either.

For the most part I do want Firaxis to make that change, succeed, and have Civ7 be a near-universally-praised successor. But I also hope that there continues to be room for competitors in the historical 4X space, an opening that has been expanded by the lack of universal satisfaction with VI, and VII's protracted development. The 4X genre is richer for this variety, even if most people have their favorites and non-favorites among the contenders.

 

[Shirotora Kenshin]

I think it's pretty unlikely we'll get a decent AI

And i must say i think the same. Still hoping that Firaxis proves me wrong… but i don‘t think they will do.

I‘m expecting the graphics in Civ7 to be overwhelming good, and this is nice, but probably they will not deliver a decent AI, i got that feeling. The weak AI made me stop playing Civ6… and this is the reason i won‘t preorder Civ7.

 

[AntSou]

Take your pick:
-Common sense logic.
-Steam statistics.
-Statements by Firaxis (if they exist).
-Comparing to other games, using statements of their devs.
-Comparing Civfanatics averages to whole community averages.
-etc

A bunch of places. As already mentioned, Steam achievements for difficulty setting skew heavily towards easier difficulties. But its also a general trend in the industry. The dev team for Hearts of Iron 4 have started that the majority of players use lower difficulty settings, for example. The simple fact is that the vast, vast majority of video game players, console and PC, are "casual" in the sense that they play games to have fun, however they define that. People who play for a challenge or to constantly improve at a game are a small minority but they are also the most vocal and tend to show up more frequently on forums and so on.

It's my fault for stupidly quoting the entire phrase. It's obvious Civfanatics will have a higher proportion of highly skilled players. My doubt about @Eagle Pursuit statement was only in regards to his assertion that the proportion of high skilled players is shrinking. I want it to be clarified what the argument here actually is. If more players are willingly engaging with Strategy games, a genre which used to be niche, then it is to be expected that the average skill will be lower. Up to this point there's nothing to argue against. However, it seems a lot of people are concluding from this that players in general are becoming more casual, which seems to me... evidently false? The floor of player competency has increased, not decreased.

There is a weird argument being made analogous to this: The population of people able to do Algebra increased over the last two centuries. The overall quality of the actual Algebra being produced decreased, evidently, since it used to be conducted by a niche of highly skilled users. We therefore conclude people are worse at Algebra today than in the 1800s.

The simple fact is that the vast, vast majority of video game players, console and PC, are "casual" in the sense that they play games to have fun, however they define that.

This definition is too vague to be useful. When talking about "casual players" it is almost always implied that these are players who wish to enjoy a game without having to deal with too much resistance or hassle. But I don't think the evidence supports that, it's a dubious argument at best.

1. @InsidiousMage How is it a trend in the industry when we are in the era of Elden Ring topping steam charts, along with its DLC which is only accessible to highly skilled users? The base game is already at 25 million copies sold.

2. Your comment about Hearts of Iron 4 skims over the fact that it is Hearts of Iron 4. Paradox Entertainment games have become more popular, not less. And they follow a DLC model with expansions that add game mechanics that further increase the complexity and granularity of the game.

3. Civilization VI leader and civ abilities are more complex than the ones designed for Civ IV. Warhammer 3 is a considerably more complex game than Shogun 1. AoE4 civ designs are more complex than any of its predecessors. In City Builders, City Skylines is a more complex game than SimCity. Anno 1800 is a more complex game than Caesar 3. You could probably look at any genre, from Fifa games to fighter games to racing games, etc, etc, and realize the baseline to be competent at those games is higher than it used to be.

4. Multiplayer is a thing. It has become more popular, not less. It is inherently competitive.

5. Roguelikes have grown in popularity.

6. @Arcaian I checked the charts, and this is the breakdown:

At least Settler - 37.3%
At least Chieftain - 33.5%
At least Warlord - 31.7%
At least Prince - 29.4%
At least King - 14.7%
At least Emperor - 10.1%
At least Immortal - 7.5%
Deity - 6.4%

This gives us the following breakdown of players winning at least once at each tier:
Settler (37.3 - 33.5) -> 3.8%
Chieftain (33.5 - 31.7) -> 1.8%
Warlord (31.7 - 29.4) -> 2.3%
Prince (29.4 - 14.7) -> 14.7%
King (14.7 - 10.1) -> 4.6%
Emperor (10.1 - 7.5) -> 2.6%
Immortal (7.5 - 6.4) -> 1.1%
Deity -> 6.4%

Resulting in:
39.4% of default difficulty achievement (14.7 / 37.3);
39.4% of higher than default difficulty achievements ((4.6+2.6+1.1+6.4) / 37.3)
21.2% of lower than default difficulty achievements ((3.8 + 1.8 + 2.3) / 37.3)

Even if you give some leeway for the 6.4% Deity wins, which likely include shenanigans, the disparity is still huge. There's a ton of people willingly engaging with higher difficulties, and more so than with the lower difficulties.
----

I'll remain sceptical about the supposed "casualization" of the player base.

I accept more people are playing games in genres that used to be niche and companies will want to cater to them. I see it as a sign of people seeking to be challenged.

Edit: @Solver you replied to me while I was writing this post.

 

[InsidiousMage]

However, it seems a lot of people are concluding from this that players in general are becoming more casual, which seems to me... evidently false? The floor of player competency has increased, not decreased.

2. Your comment about Hearts of Iron 4 skims over the fact that it is Hearts of Iron 4. Paradox Entertainment games have become more popular, not less. And they follow a DLC model with expansions that add game mechanics that further increase the complexity and granularity of the game.

3. Civilization VI leader and civ abilities are more complex than the ones designed for Civ IV. Warhammer 3 is a considerably more complex game than Shogun 1. AoE4 civ designs are more complex than any of its predecessors. In City Builders, City Skylines is a more complex game than SimCity. Anno 1800 is a more complex game than Caesar 3. You could probably look at any genre, from Fifa games to fighter games to racing games, etc, etc, and realize the baseline to be competent at those games is higher than it used to be.

Player competency and game complexity don't have anything to do with how people play games. Again, most people play HoI4 on lower difficulty levels, meaning the play in away that is less punishing to player mistakes. Same thing with Civ6, you can make far, far more mistakes on lower difficulty levels than higher. To me, this means there is a certainly level of "getting good" that most players are willing to reach and then concentrate on having fun over improving their skill at the game ie. playing casually.

Also, I'm sure you could an incredibly strong argument that the increase in player competency could almost entirely be attributed to YouTube and the internet generally. The EU4 wiki has entire sections for nations devoted entirely to what kind of strategy use and you can find in depths for every game on YouTube. It would be shocking if the general level of player competency didn't increase given how much information is out there.

1. @InsidiousMage How is it a trend in the industry when we are in the era of Elden Ring topping steam charts, along with its DLC which is only accessible to highly skilled users? The base game is already at 25 million copies sold.

First off, Elden Ring would obviously be an exception to the rule, which always exist. Second, Elden Ring is notably different than other FromSoft games in that it is an open world game which means that you go do something else in game if you get stuck on a boss, which you can't always do in other FromSoft games. There is also just a lot more to do in Elden Ring compared to every other FromSoft game as well meaning players have a lot more opportunities to improve and level up without having to constantly refight the same boss over and over. Finally, There is also the fact that Elden Ring is considered notably easier than other FromSoft games and there was a huge controversy about how much more difficulty the DLC was compared to the base game.

4. Multiplayer is a thing. It has become more popular, not less. It is inherently competitive.

But has it become more popular relative to the number of people actually playing? Competitive multiplayer can see an increase in popularity while actually become a small part of the playerbase overall. The increase in popularity of any given type of play is only meaningful in relation to the others. Saying competitive multiplayer has gotten more popular doesn't tell us anything that useful without know how popular other types of play have gotten.

 

[Leyrann]

There is a weird argument being made analogous to this: The population of people able to do Algebra increased over the last two centuries. The overall quality of the actual Algebra being produced decreased, evidently, since it used to be conducted by a niche of highly skilled users. We therefore conclude people are worse at Algebra today than in the 1800s.

Perhaps not the 1800s, but considering Euler lived in the 1700s you could certainly argue the quality of algebra being produced has decreased across the past 250 years.

 

[Arent11]

This is irrelevant. Aside from that, you and me as consumers and players do not want a good AI. Studies show we automatically think a good AI is unfair and/or cheating.

This People want to play against waves of stupid AI & stop them with a single phalanx at Termophilae

There could be an *additional* option for a trained AI at deity level. This would be interesting for experts.

 

[GeneralZift]

I don't understand how this is even a discussion. What is there to discuss? The AI sucks and people want a better AI. I thought this was both common sense AND common knowledge.

 

[DeckerdJames]

Depends upon how they do difficulty levels. If AI competence scales with difficulty level instead of relying on massive bonuses, then it might work out?

That is an important point. The difficulty in Civ 6 isn’t only strategic difficulty. Many people do not rate the Civ 6 AI as being very smart in a strategic way. Instead, the difficulty is in the speed of the race. All of the victory conditions are a type of race. This is an intrinsic quality of a 4x game. Time is an important factor. Deity difficulty gets bonuses that help it reach those 4x goals more quickly. Now, that does require the player to pursue a strategy related to efficiency, but it dominates the game on high difficulty settings. If the players are on even ground in terms of 4x advancement, a player might be able to have greater flexibility in their strategic goals because a great deal of emphasis could be taken off of pure speed of advancement.

I see people often talk about how the AI doesn’t do much militarily in the middle and late game and cite that as a lack of intelligence, but production invested in military goals is production taken away from 4x advancement goals. Producing 10 builders might more often move the AI closer to victory than producing 10 military units would. Producing military units is more about depriving the enemy of something, but what good is it to take a couple of cities from your closest neighbors in the late game when someone on the other side of the world puts all its effort into a 4x victory condition and wins the game.

Of course having sufficient 4x advancement will still be a consideration even without 4x bonuses applied to the AI, but without the 4x bonuses there will be more time to convert military successes into total victory so military goals would be more viable and frequent.

It’s worth pointing out that a human player and a computer player need to be very good judges of how long it takes to do things. They have to be a good judge of time.

 

[Gedemon]

Sorry, but this is like that video about some guy asking for WoW Classic and getting the reply "you think you want it, but you don't".

No, I'm pretty sure I and many others do want good AI.

Thing is we all have a different idea of what a "good AI".

I mean, sure, you want a good AI, I want a good AI, and Firaxis want to code a good AI.

But in the end, those may be three very different AI.

For an extreme example from past discussions, I remember that what we'd expect from a good "RP" IA is quite different of what we'd expect from a good "competitive" AI, that both seem incompatible on some key design elements, to the point that RP players would not want to play against a "good" competitive AI, and competitive players would not want to play with a "good" RP AI.

That said I do hope that we'll get a decent AI this time, my definition of it being an AI that is able to use all of the game's features.

Now, if you ask what is my definition of a "good AI", it's a short one: an AI that is fully moddable.

 

[King Flevance]

For me, a good AI is not just simply "good at exploiting the game's mechanics for a challenge."

I personally would like the AI to be more modelled toward atmosphere and personality. When I play civ, I prefer to sort of roleplay. I like Civ for its simplistic design but flexible strategies. Various winning conditions has become a staple in civ with diplomacy and culture offering non-wargame options. I do want the AI to be capable of exploiting the map and fielding an army well, but if I want a difficult 4x wargame, there are better options than Civ. I like Civ's charm that I could become best friends with my neighbor all game long without the worry of them betraying me - necessarily. I have even voted for an AI to win Civ 4 in diplomacy. I do not like AI to have erratic, video-gamey anti-player behavior. I want them to play the game and not be naive but also value relations with AI and players equally.

However, tastes vary.

 

[AntSou]

Player competency and game complexity don't have anything to do with how people play games. Again, most people play HoI4 on lower difficulty levels, meaning the play in away that is less punishing to player mistakes. Same thing with Civ6, you can make far, far more mistakes on lower difficulty levels than higher. To me, this means there is a certainly level of "getting good" that most players are willing to reach and then concentrate on having fun over improving their skill at the game ie. playing casually.

But I don't see how this is relevant? What is this how you're talking about and how does it relate to the level of difficulty people are willing to engage with?

It doesn't make a lot of sense to me to argue playing on lower Hoi4 difficulties is a sign of being a casual player, i.e., seeking less of a challenge, when playing Hoi4 in itself represents the opposite of that, regardless of what difficulty a player selects. The difficulty level is a detail overshadowed by the game in question being played. Any player playing Hoi4 is not a casual player and is not looking for a casual strategy experience. All paradox strategy titles require quite a bit of commitment and skill, even CK3.

I don't want to get distracted from the point and bogged down in a discussion about what a "casual" is or isn't. We're talking about AI in the context of how much difficulty people are willing to engage with. Pointing to Hoi4 as an example of people seeking casual gaming experiences is a contradiction.

Also, I'm sure you could an incredibly strong argument that the increase in player competency could almost entirely be attributed to YouTube and the internet generally. The EU4 wiki has entire sections for nations devoted entirely to what kind of strategy use and you can find in depths for every game on YouTube. It would be shocking if the general level of player competency didn't increase given how much information is out there.

Sure, but this also isn't relevant? Reading and watching tutorials is a time commitment. Going to such lengths to learn a strategy game is also contrary to a casual attitude towards gaming.

First off, Elden Ring would obviously be an exception to the rule, which always exist.

Well, it needs to "obviously be an exception to the rule", otherwise your argument is weak. But you casually dismiss it as self evident and immediately move on, not explaining why we should dismiss it as an exception.

Second, Elden Ring is notably different than other FromSoft games in that it is an open world game which means that you go do something else in game if you get stuck on a boss, which you can't always do in other FromSoft games. There is also just a lot more to do in Elden Ring compared to every other FromSoft game as well meaning players have a lot more opportunities to improve and level up without having to constantly refight the same boss over and over. Finally, There is also the fact that Elden Ring is considered notably easier than other FromSoft games and there was a huge controversy about how much more difficulty the DLC was compared to the base game.

You're doing the same thing you did with the Hoi4 example. Elden Ring is a difficult game and the most popular FromSoft title. Pointing to other FromSoft titles existing doesn't prove anything, if anything it does the opposite.

But has it become more popular relative to the number of people actually playing? Competitive multiplayer can see an increase in popularity while actually become a small part of the playerbase overall. The increase in popularity of any given type of play is only meaningful in relation to the others. Saying competitive multiplayer has gotten more popular doesn't tell us anything that useful without know how popular other types of play have gotten.

Ok, sure.
----

I feel like we're talking too much, honestly.

Civ VI AI was clearly deficient in many ways. The steam achievements show people were happy with either playing with default settings or seeking a bit more challenge. Improving AI competency is not going to make the game unplayable to the average user.

Statements along the lines of "Firaxis will want to appeal to their core player base, not the low amount of highly skilled users" may be self-evident but not contrary to having an improved AI in the game, and it's condescending.