civ 6 is a good game with issues.

[Zegangani]

By playing them, it's not "great" anymore, it's one level above at least

You always making me want to try Civ 4 (and it's mods), especially since I'm discouraged to play Civ6 untill it gets a BugFix Update. I might do that sometime.

The only thing that prevents me from doing that is my (small) hope that Civ6 might get that update till at least Fall 2021, and making some mods for Civ6 (yes modding without actually playing the Game lol) before that hope is drained.

 

[Kupe Navigator]

This fellow Kiwi makes a quite a good point on why people think civ 6 sucks and He gives fair points. I agree 100% with him.

edit: here is his review of all civ 6 which is fair and balanced

it is just nice to hear from someone who recognizes the problem with civ 6 BUT not going "RUINED FOREVER!!!! DEVS SHOULD BE ASHAMED OF THEMSELVES!"

Moderator Action: Post edited. Please discuss the topic not the forum member. ~ LK

 

[Socrates99]

I have to disagree here. Having different rules for the AI would make the game not understandable and overly complex. "Cheating" is in no way an acceptable path to patch an incapable AI. But on the other hand, saving / loading is not acceptable either, I never do that, this is plain cheating too.

To me, a strategy game needs to have plain and clear rules, and all players abide by those same rules. I cannot imagine enjoying a game where one team plays soccer (no hands allowed) against another team playing rugby. A plain mess if you ask me.

AI could be improved. The devs are showing us that by simply doing the opposite: every new patch seems to make it worse, either because the newly introduced mechanics are clearly disregarded, either with apparent typos (see the "AI crazy about science" thread), and insufficient playtesting. @Infixo is apparently having some success in improving things with what is made available to modders, so it is possible. Deciding on your goals and pursuing a victory in a consistent way seems to be a key requirement that the AI is not doing, for instance.

Name a game where AI doesn't "cheat" or have a major advantage on higher difficulty. I'm not saying asking for improvements is unreasonable but you'd be hard pressed to find anything even minutely complicated where AI is competitive with a halfassed competent human.

 

[Alexander's Hetaroi]

Yeah.... with korean leader who looks Chinese AND lives in a palace that is built by a guy who massacred his descendants.

Historical accuracy seems to take a backseat when discussing how great a game is, focusing more on actual gameplay. Even if the gameplay isn't the greatest, I still think this is the greatest in terms of choices of civs over the choices of others from previous games. The most glaring examples are Maori over Polynesia and Gaul over the Celts.

 

[Heyesey]

Historical accuracy seems to take a backseat when discussing how great a game is, focusing more on actual gameplay.

Well if we're going to demand historical accuracy, you die after the first two turns and the game is over. Not many people lived more than 50 years back then, let alone six thousand ....

 

[Alaindor]

Name a game where AI doesn't "cheat" or have a major advantage on higher difficulty. I'm not saying asking for improvements is unreasonable but you'd be hard pressed to find anything even minutely complicated where AI is competitive with a halfassed competent human.

Chess. Go.
... but I suppose you were referring to 4X type games?

Humans can get an overall vision and take risks or gamble away, which is probably more difficult to program for the AI (... I'm not convinced though...). But what the AI can do better (if programmed properly) is make no mistales in min/maxing. I always make some occasional errors when I forget to check for the GP points, move a governor, forget to switch priorities in a city... An "AI" would not do that. So that should be an advantage to it. It may not compensate, but in early game it can be crucial, and winning this game happens in the early stages.

When I see at work what we can do with a real AI and tons of data, I'm sure we could use similar techniques for Civ. How CPU-intensive that would be might make it difficult to port to consoles though, if one wants an acceptable time between turns. Then it could use a central computing server where calculations are made, and data transferred to and fro the game running on our end-user devices. But that would be a different business model, probably not one Firaxis would want to go through.

 

[DizzKneeLand33]

Chess. Go.

When Deep Blue played Kasparov, iirc, it was programmed with an entire openings database -- so, indeed, it was a cheat, since humans in over-the-board competitions were not allowed to access the same database/opening books (in correspondence they were, but this was an over-the-board match). Go was different, but that was 20+ years later.

All I'm saying is that no one really complained about the "cheats" back then, they just wanted a competent AI. Without that openings programming, the chess computers of the 90's would not have been competitive in an way versus top GM's.

 

[UWHabs]

When Deep Blue played Kasparov, iirc, it was programmed with an entire openings database -- so, indeed, it was a cheat, since humans in over-the-board competitions were not allowed to access the same database/opening books (in correspondence they were, but this was an over-the-board match). Go was different, but that was 20+ years later.

All I'm saying is that no one really complained about the "cheats" back then, they just wanted a competent AI. Without that openings programming, the chess computers of the 90's would not have been competitive in an way versus top GM's.

I mean, technically someone could study and memorize all the openings possible, so in that sense the AI did not "cheat". But yeah, obviously it had access to a lot more knowledge and memory than any human could possibly imagine. It would be like in civ, if the AI saw your unit on a tile, and then remembered all the possible tiles that you could reach in X moves - a human would technically be able to do that, but not to the level that a computer could.

But yeah, where civ AI comes out, it's a tough question. As @Alaindor mentioned, you could offload civ computing to some cloud system to truly optimize the bejeezus out of things, but the computing power to handle that obviously would never make sense for a game like civ unless if you're charging people 10 bucks an hour to play or something. Okay, maybe you could do like 50 cents or 1$ an hour, but to have enough computing power to do anything of note, you're probably needing to go higher to cover those costs. I hope that the AI is at least optimized for the "basic" items in its control - stuff like policy cards, tile management, etc... things that have virtually no connection to outside factors. Those are pieces where any simple valuation/weighting system can pretty easily handle what to run, and should be pretty much as good as a human, other than some human look-ahead pieces - ie. I know that my 50% off upgrades card is coming up soon, so maybe I decide to run a +money card earlier on to build up the cash to handle them, whereas an AI may not quite see that. Or maybe if I know that this current policy selection will only last for the next 3 turns, I make a different choice on what to run vs thinking it might be 10+ turns until my next switch.

But in terms of the AI "cheating", it always comes down to what exactly it does to cheat to figure out whether it makes sense. Generally speaking, if you can see the AI cheating, it's probably not right. So, for example, if all AI units got an extra movement, or started with a free promotion, or an extra builder charge, I think we would be rioting. But if the AI gets free policy changes whenever they feel like it? I doubt any of us would be able to to notice that without digging into the logs.

 

[Socrates99]

Chess. Go.
... but I suppose you were referring to 4X type games?

Humans can get an overall vision and take risks or gamble away, which is probably more difficult to program for the AI (... I'm not convinced though...). But what the AI can do better (if programmed properly) is make no mistales in min/maxing. I always make some occasional errors when I forget to check for the GP points, move a governor, forget to switch priorities in a city... An "AI" would not do that. So that should be an advantage to it. It may not compensate, but in early game it can be crucial, and winning this game happens in the early stages.

When I see at work what we can do with a real AI and tons of data, I'm sure we could use similar techniques for Civ. How CPU-intensive that would be might make it difficult to port to consoles though, if one wants an acceptable time between turns. Then it could use a central computing server where calculations are made, and data transferred to and fro the game running on our end-user devices. But that would be a different business model, probably not one Firaxis would want to go through.

Eh, I didn't necessarily mean 4x. Platformers, rpgs, sports games, racing games, etc. The computer gets major advantages once the difficulty gets high enough on almost any game. Once the player recognizes the patterns, weaknesses, etc the only way to stop that is to buff the AI and wait for the player to mess up.

Chess is simple enough that an AI can be programmed with human strategies like Dizz points out. Take chess and give half the pawns bows, add variable terrain to the board, give the pieces hit points rather than one shot one kill, add a tech tree to the game that changes the properties of the pieces as the game progresses, add spawn points where units can be created from resources you collect and you'd find AIs need a boost to compete again. It'd take away the standard strategies and create the need for improvisation and that's where our brains win. If we're on even footing that is.

I'm just saying I'm not against defeating a stronger but dumber AI. It's what I've been doing and enjoying for 30 years in almost every video game I've played.

I do agree some optimization would be good. Civ VI AI severely lacks focus. It should be tuned t relentlessly pursue one or two VCs and ignore the rest. Know what I hate more than not improving tiles or stupid unit moves? Watching Genghis Khan pour thousands of gold into emergencies while spamming missionaries. Or watching Seonduk build a massive aluminum demanding military and letting it sit while floundering in her SV. Watching the AI attempt all VCs at once drives me crazy. Just flavoring it to ignore wasteful resource spending would improve it by magnitudes.

 

[MrRadar]

Or watching Seonduk build a massive aluminum demanding military and letting it sit while floundering in her SV.

AI will just not research Plastics and go further that line until very late, therefore this standstill after the Moon landing. And this is very consistent across the games and AIs.

Watching the AI attempt all VCs at once drives me crazy.

At this point I don't believe the AI "attempts" anything. It is just a few buckets filling up at random, and the one that happens to fill up first, wins the AI the game. If there are some decision making mechanisms for the AI to prioritize some of those buckets, they fail miserably.
This game now is a SimCiv game, where the player can very safely achieve a routine victory at any difficulty.

 

[Socrates99]

AI will just not research Plastics and go further that line until very late, therefore this standstill after the Moon landing. And this is very consistent across the games and AIs.


At this point I don't believe the AI "attempts" anything. It is just a few buckets filling up at random, and the one that happens to fill up first, wins the AI the game. If there are some decision making mechanisms for the AI to prioritize some of those buckets, they fail miserably.
This game now is a SimCiv game, where the player can very safely achieve a routine victory at any difficulty.

Kind of a Tom-ay-toes vs Tom-ah-toes point really. Trying to fill all buckets at once to see what fills first is going for all VCs.

 

[Aristos]

This game now is a SimCiv game, where the player can very safely achieve a routine victory at any difficulty.

I assume you meant "this franchise"? This game (civ 6) was a SimCiv since the very beginning.

And yes, that is exactly my impression; the code is so poor that the general behavior of the AI is to just fill buckets to, in the mind of the developers, give some impression of "playing like a human"... what I cannot discern, though, is if the devs are such bad players of their own game that they think that all players "just fill buckets and therefore our AI plays like a human", or if they are just less-than-optimal coders...

The third option, of course, and one that I never disregarded, is that the devs/suits are so smart that they know the majority of the (new?) audience "just fills buckets" and wants to just "safely achieve a routine victory at any difficulty"...

 

[Zegangani]

I don't think it's that difficult to make Civ6's AI better, eg. smarter (knowing how every game systeme work and how they are connected with each other, setting priorities, better unit management, perhaps some short/medium-term decisions (like prioritizing building mili Units when they plan starting a war) ...etc). The only downside of an AI that is smart and competitive, is the amount of time per turn for an AI Civ that it would need to process all it's in-game knowledge of things to set priorities and make decisions.

We all take our Time to do things in our Turn (not necessarily in a peaceful late game where you just press next turn), but the AI would also need its time to make its decisions (assuming we have that smart AI), and count that for the many civs in the Game, the waiting Time between turns for the Human Player could be very long.

 

[Zaarin]

Well if we're going to demand historical accuracy, you die after the first two turns and the game is over. Not many people lived more than 50 years back then, let alone six thousand ....

You're not playing as a person; you're playing as a civilization. You still have a point, though, insofar as not many civilizations have lasted more than a couple thousand years.

 

[PiR]

We all take our Time to do things in our Turn (not necessarily in a peaceful late game where you just press next turn), but the AI would also need its time to make its decisions (assuming we have that smart AI), and count that for the many civs in the Game, the waiting Time between turns for the Human Player could be very long.

As it was in previous editions. That's why I believe the state of the AI is due to that optimization of turn time.

 

[Zegangani]

As it was in previous editions. That's why I believe the state of the AI is due to that optimization of turn time.

Yes, me too. Modders who add long lua scripts (that process a lot of data) to the Game, are always afraid of making the Game slower because of that. Yes, scripts need to cross an extra door to get to the Core Game's Engine, hence need more processing time, but even changes made directly into the DLL would make for a longer waiting time than in the usual civ games.

What are People's opinion about the waiting time between turns in vanilla civ5 and civ5 Vox Populli? I assume it's a bit longer.

 

[PiR]

It's also longer and longer the more the game lasts, with more units, more cities, etc. The early game is often fine.

 

[Alexander's Hetaroi]

Well if we're going to demand historical accuracy, you die after the first two turns and the game is over. Not many people lived more than 50 years back then, let alone six thousand ....

More like after the first turn considering it can last several hundred years.
I'm not looking for historical precision though, and that's one aspect where the series can't be totally accurate. I do demand less zombies in the next iteration though.

You're not playing as a person; you're playing as a civilization. You still have a point, though, insofar as not many civilizations have lasted more than a couple thousand years.

I'm assuming the statement was regards to always playing as an immortal leader, not just playing a civilization.

 

[aieeegrunt]

AI will just not research Plastics and go further that line until very late, therefore this standstill after the Moon landing. And this is very consistent across the games and AIs.


At this point I don't believe the AI "attempts" anything. It is just a few buckets filling up at random, and the one that happens to fill up first, wins the AI the game. If there are some decision making mechanisms for the AI to prioritize some of those buckets, they fail miserably.
This game now is a SimCiv game, where the player can very safely achieve a routine victory at any difficulty.

SimCiv is the perfect analogy

I don't think it's that difficult to make Civ6's AI better, eg. smarter (knowing how every game systeme work and how they are connected with each other, setting priorities, better unit management, perhaps some short/medium-term decisions (like prioritizing building mili Units when they plan starting a war) ...etc). The only downside of an AI that is smart and competitive, is the amount of time per turn for an AI Civ that it would need to process all it's in-game knowledge of things to set priorities and make decisions.

We all take our Time to do things in our Turn (not necessarily in a peaceful late game where you just press next turn), but the AI would also need its time to make its decisions (assuming we have that smart AI), and count that for the many civs in the Game, the waiting Time between turns for the Human Player could be very long.

It’s not difficult at all to make this AI better, look at any of the AI improving mods and remember that they are limited to parameter manipulation and some scrupting

Hell look at the AI Crazy About Science thread

Civ6 is an excellent game for people who like a peaceful building style, and I have to assume that was done deliberatly

 

[Aristos]

I don't think it's that difficult to make Civ6's AI better, eg. smarter (knowing how every game systeme work and how they are connected with each other, setting priorities, better unit management, perhaps some short/medium-term decisions (like prioritizing building mili Units when they plan starting a war) ...etc). The only downside of an AI that is smart and competitive, is the amount of time per turn for an AI Civ that it would need to process all it's in-game knowledge of things to set priorities and make decisions.

We all take our Time to do things in our Turn (not necessarily in a peaceful late game where you just press next turn), but the AI would also need its time to make its decisions (assuming we have that smart AI), and count that for the many civs in the Game, the waiting Time between turns for the Human Player could be very long.

Multithreading.

It has been done before with success. Galciv AI calculates most of their turn during the human player's turn. Then, it only adjusts what has changed during human turn. Result: a much better, and much faster AI.