Complex game mechanics or good AI?

[Aristos]

That's a strategy, not a set of rules.

If it ALWAYS works, it's not a strategy.

 

[PiR]

If it is always works, it's an efficient strategy, still not a set of rules.

 

[aieeegrunt]

If it ALWAYS works, it's not a strategy.

That was my point. Complicated does not mean deep.

Removing buildings from the city centre was an interesting idea, but the execution was straight up unneccessarily complicated with the whole district concept and it’s silly restrictions.

Merge the concept of improvements and districts. Then your population limit is actually a simple mathematical limit rather than some arbitrary tier thing.

Put a library on a tile. Assign a pop to it. Or a mine, or a farm. Simple, clean, efficient.

This also balances wide verses tall. Spam a bunch of low pop cities and congratulations, you are pre-Peter the Great Russia.

 

[oSiyeza]

This is a false dichotomy.

First if the AI does not handle the game mechanics is not a good AI. Its a bad AI that you dont notice is bad because the game is so simplistic the player does not notice.

Second, AIs can handle complex mechanics. It just happen that sometimes AIs cant handle mechanics that are designed with complete disregard of the AI that needs to play them.

Third, Civ VI for the most part does not have complex mechanics. Pathfinding, trade, citybuilding, improvement reparing, using nukes, using planes, using carriers, trying to win at a selected victory path... these are not complex mechanics. Most notably the new game modes do not add complex mechanics; the AI is bad at handling these mechanics because Fxs did not care about an AI capable of handling them. And because they kept adding new mechanics faster than they could fix the AI.

Finally, there are things in Civ VI that are objectively very complex and difficult to do, and because of that the AI is incapable of doing them. Like conquering cities. And here the choice is also not about complex mechanics and a good AI. Because things like taking cities are not just hard and complex, but needlessly hard and complex. Many of the mechanics the AI has problems with are just bad mechanics that nobody cared to test with the AI and say: "you know what, this does not work, lets think it a bit more".

So no, the only real answer is we deserve good mechanics and a good AI, developed together, with a proper planing.

 

[aieeegrunt]

This is a false dichotomy.

First if the AI does not handle the game mechanics is not a good AI. Its a bad AI that you dont notice is bad because the game is so simplistic the player does not notice.

Second, AIs can handle complex mechanics. It just happen that sometimes AIs cant handle mechanics that are designed with complete disregard of the AI that needs to play them.

Third, Civ VI for the most part does not have complex mechanics. Pathfinding, trade, citybuilding, improvement reparing, using nukes, using planes, using carriers, trying to win at a selected victory path... these are not complex mechanics. Most notably the new game modes do not add complex mechanics; the AI is bad at handling these mechanics because Fxs did not care about an AI capable of handling them. And because they kept adding new mechanics faster than they could fix the AI.

Finally, there are things in Civ VI that are objectively very complex and difficult to do, and because of that the AI is incapable of doing them. Like conquering cities. And here the choice is also not about complex mechanics and a good AI. Because things like taking cities are not just hard and complex, but needlessly hard and complex. Many of the mechanics the AI has problems with are just bad mechanics that nobody cared to test with the AI and say: "you know what, this does not work, lets think it a bit more".

So no, the only real answer is we deserve good mechanics and a good AI, developed together, with a proper planing.

I totally agree; Civ 6 is complicated and fiddly

 

[Aristos]

If it is always works, it's an efficient strategy, still not a set of rules.

Not if it always works because the AI is not capable of, at minimum, playing its own game using ALL mechanics available.

 

[uhu]

If you replace "complex game mechanics" with "fun AI" you can find a lot of old discussion to this contrasting. The keyword is here asymmetric which Gedemon already introduced in the thread. Short: The humans side is far more complex and features also gimmicks which many players like. The AI players play a simpler game. Both play by significantly different rules and that is ok. I don't see that as "cheating". There is never something like a "level playing field" or such. We humans make the game rules in ways we like it and we are good in playing. Rules in ways which benefit computers are "No fun" for us. As a result we insist in playing "our" game and bemoan that AI players are weak in doing that. Don't get me started about "fairness and cheating"!

Not if it always works because the AI is not capable of, at minimum, playing its own game using ALL mechanics available.

"AI is capable of" requires any code in the first place. Some parts look like there is code still missing - cultists, special abilities of heroes, in vanilla aircraft ... yeah, programming is debugging empty files

I think, many of us don't want to indicate on others - but just make the game better and backfill the mentioned holes as well as try to implement some more AI improvements.
Kevin mentioned recently an older video in which dshirk emphasized in context of the World Builder "We are not done!" - so what? Is the World Builder in a appropriate shape now?
We need: One more (decisive) blow!
(already existing content in a Goody Hut, ok?)

 

[ggalindo001]

Interesting thread, and one I hope the devs are looking at.

1. Asymmetric is the right answer here. I know players don't want the CPU to "cheat", until they "do".
2. I would like to see more variability in the difficulty settings vs. just simply providing more yields to the CPU. As you ratchet up the difficulty levels, AI aggressiveness goes up, and more naturally aggressive AI are even stronger. And they cheat a little bit as well (and more and more). It is harder to make deals with AI, and harder to make peace, etc., AI more apt to nuke you, etc.
3. Limited multiple units per tile -- what Humankind is doing is interesting, I also like one of the Mods that allow for stacking of melee, cav and ranged into one tile. The AI seems to do a lot better in this regard.
4. Allow for settings that handicap human players only. I made a mod where I limited the number of alliances to just 2 -- but it applies to all players -- I would ideally like that to only apply to the human player.
5. Allow for settings that only apply AGAINST human players only -- so war weariness, etc., is not evaluated AI vs. AI, but only against the human player.

 

[Melkus]

A lot of interesting points here, but it feels to me like we're letting Firaxis off the hook again. (This is not meant as some kind of savage criticism, just trying to put things in perspective as history gets forgotten so quickly nowadays.)

1) When Soren Johnson led CIV 4, Firaxis' thinking on these lines was much further evolved. They even integrated parts of the best AI mod at the time in "Beyond the sword". (And threw in a version of Rhye's and Fall, as well!) I realize that modding is not possible at this level yet, but couldn't they have learned something from Vox Populi (like, tactical unit movement in 1 UPT or achieving a balance in city sieging mechanics that the AI can work with)?

I mean, at the time Google invited the man to explain cutting edge game design (with special regard to AI)

The current dev team is super likeable and creative, but no one there seems to have a similar ability to reflect on what they're doing. Many of their public statements seem a bit naive, really. (Yes, I realize that we don't have access to internal discussions).
They put a board game designer in charge, and he has designed a great board game (which inherently leans towards Multiplayer), but this other dimension of computer gaming is missing a bit.

2) The current AI has clear weaknesses which can be seen easily when you watch an AI only game (or just play the Cree and ally a few of them). This is not a matter of developing ingenious algorithms or employing machine learning. This is a matter of shifting a modest amount of resources to the improvement of the AI.

Also, to the original question - and this false dichotomy has been pointed out - the complexity of the game can also be raised by improving the AI (and raising the diversity of their approaches). In fact, I would argue that this may be the easiest way to do so at this point, though not one that generates a lot of money.
And piling on mechanics without telling the AI how to work them decreases complexity.

 

[EgonSpengler]

A variation on "Why can't I have both?": I pick whichever option would go furthest in addressing the late-game malaise. Civ VI stalls and becomes terribly boring once my Civ has reached a certain level. As mentioned already, success begets success, so once I get ahead in the foot-race, I'll always stay ahead. Once I'm winning, I've won. The whole rest of the game is just pressing 'End Turn' and doing maintenance, like upgrading units and renewing trade deals when they expire. I don't think I've formally finished a game of Civ VI in 5 years. Neither of these options would fix this problem by itself, but I could see either of them being a useful part of a correction. I choose whichever one would make the second half of the game worth playing.

(Before anyone mentions it, raising the difficulty setting doesn't change this, because it doesn't change the game's mechanics or improve the AI. Raising the difficulty increases the possibility that I'll get rolled in the Ancient or Classical Eras, and it delays the moment I pull ahead for good. In those ways, raising the difficulty makes the early parts of the game more fraught and exciting, but that's not the part of the game that needs help.)

If it is always works, it's an efficient strategy, still not a set of rules.

You say 'efficient', I say 'rote.' I agree with your point, though, strategies and rules are different. The best rules sets enable myriad strategies, from which players would choose depending on circumstances.

 

[PiR]

Try a Modern era start. Early part of the game = late-game.

 

[Leucarum]

Reducing mid/late game micro would go a long way to making the game drag less.

Overhauling religion (the biggest clickfest ever) and maybe removing builders in favour of purchasing tile improvements directly could both help.

 

[aieeegrunt]

Reducing mid/late game micro would go a long way to making the game drag less.

Overhauling religion (the biggest clickfest ever) and maybe removing builders in favour of purchasing tile improvements directly could both help.

I would love both ideas