Complex game mechanics or good AI?

[ezzlar]

The amount of content that has been added to Civ6 is in a way truly amazing. We have a number of civs with unique abilities, buildings, districts and infrastructure. The different game modes. And of course the two expansions.

But it is also somewhat obvious that the AI really can´t handle all the fine aspects of the additions. Not by themselves and certainly not mixed together for synergy. The AI is even struggling with district planning which is kinda basic (and then there is this thing about what somebody labelled the sliding minigame puzzle movement challenge).

However, instead of another "Do you like the AI?" thread I have been thinking about two different directions the game may evolve into. The first one is with all the goodies and mechanisms. This will probably be a sandbox game where the AI opponents are there more for show and immersion. Here it is up to you to decide what the goal of the game is, roleplaying, SimCiv, yield porn or whatever. What the game lacks here is internal threats. The Rhye mechanic of stability is one way of handling this which is quite impressive. In Civ6 success breeds more success and the famous snowballing effect. Since a lot of players struggle with the game, rubberbanding or negative modifiers are usually frown upon. But I think they would need to be added to offer a "challenge yourself" game.

The other alternative is of course to scale down all the complexities and develop new features always with the limited AI in mind. By doing this your AI opponents would clearly be the main obstacle. A lot of other problems would disappear this way. The vox populi mod of Civ5 clearly makes the AI something to consider.

So what would you prefer? I don´t see how the devs could create an AI that handles all the additions to the game in a way which the playerbase would regards as competent. It would indeed be an amazing feat if they did but nothing worth hoping for this late in the game lifetime cycle.

 

[Linklite]

Your first option seems to be more of an attitude change towards the game rather than the game itself. That's fine, but I largely see the game as that anyway - I don't view the AI as an opponent so much as I see them in the way that I see CSs or Barbs, obstacles (or helps) while I try to grow and improve my empire. Essentially, as I see this suggestion, it's the equivalent of "suck it up and git gud, you need to change not the game". There is some validity to this, but I doubt it would be popular.

On your second point. You can't scale down the game. The entire point of a 4X is the complexity, the various different choices are important. There are some minor tweaks you can make that might improve things, for example it was suggested that 1 UPT doesn't apply to civs and CSs that youbare not at war with, but most of it is an important part of the game that its absence would detract from the game. The inability to plan out your city would make the game much simpler and while the AI could probably cope with better, the game itself would be less fun without it.

I think that Civ 6 crossed the Rubicon with its complexity. It cannot go back because the game itself would feel like a regression. Asking the player base to reconsider how it views the game isn't going to work, they're too set in their ways. I think if Firaxis wants to resolve this, they'll just have to fork out the money and redesign the AI for VII.

 

[Vilvatiktan]

To simplify the options, isn't it safe to conclude that:

Complex game mechanics are targeted for more multiplayer gameplay as each players could exploit one of the many game mechanics in favour of their victory whilst the AI would not be as smart as the human player to exploit or even utilise any of those complex game mechanics, leading to single-player games feel bland and players quitting the games early as we have seen in Civ6.

As for Good AI would target single-player games (also multiplayer games) as the AI can make each game feel more immersive as the AI can utilise the mechanics no matter how many or how few are them. This can also improve multiplayer games with players can trust that the AIs wouldn't be so lame if any player decided to leave the game at any point during the game. The problem: It could be expensive for the devs to develop good AI, isn't it? )

 

[Noble Zarkon]

Your first option seems to be more of an attitude change towards the game rather than the game itself. That's fine, but I largely see the game as that anyway - I don't view the AI as an opponent so much as I see them in the way that I see CSs or Barbs, obstacles (or helps) while I try to grow and improve my empire.

I prefer the sandbox environment myself, internal threats come from the Loyalty mechanism and Apocalypse Mode (if switched on) - can you win before the end of the world!!

Complex game mechanics are targeted for more multiplayer gameplay

Totally disagree - complex mechanics are what makes single player an enjoyable challenge whereas from what I have seen multi-player is just a war game loosely based around Civ where half the mechanics are banned or modded out !!

 

[Alaindor]

I'm sure better AI can be developed but using more modern techniques and machine-learning : take a holistic approach and log / analyze all the real games played (uploading anonymous detailed stats over time) and the AI will get smarter over time - possibly, unbeatable unless player gets very lucky.

I am sure this more modern approach would be better suited than the current if-then static set of rules.

Complexity is what makes the game interesting: please do not degrade it!

 

[uhu]

Totally disagree - complex mechanics are what makes single player an enjoyable challenge whereas from what I have seen multi-player is just a war game loosely based around Civ where half the mechanics are banned or modded out !!

If it didn't exist, one might think it is impossible: on one side the epic theater of whole human history from the beginning of awakening until the "end of time" and on the other tiny episodes of multiple humans in their desire for rapid decisive blows like in Red death.

I think that Civ 6 crossed the Rubicon with its complexity. It cannot go back because the game itself would feel like a regression.

Not just Civ 6, but the whole franchise. Sid's 3x 1/3 are a good goal for smart designers - still marketing and customers demand forever growth.

Asking the player base to reconsider how it views the game isn't going to work, they're too set in their ways.

The vast majority of discussions about altered gameplay and mechanics finds very quickly very complex and complicated proposals because we like that. Solutions which would help the AI are typically rated unsexy and No Fun! - if consideration is given to AI aspects in the first place. Dogmatic cats.

The other alternative is of course to scale down all the complexities and develop new features always with the limited AI in mind.

I would wish the developers could lay emphasis on AI limitations at least during the design of details.

I don´t see how the devs could create an AI that handles all the additions to the game in a way which the playerbase would regards as competent.

I think if Firaxis wants to resolve this, they'll just have to fork out the money and redesign the AI for VII.

It could be expensive for the devs to develop good AI, isn't it? )

Doesn't it look like there is a pink hippo_potamuse in the room above the cupboard?!

Reminds me of Aron Bodenheimer's booklet:

Why? About the obscenity of asking.

 

[Kaan Boztepe]

we always talk in these forums about AI when i am pretty sure there is no AI in the game at all. The Devs have made a table with numbers associated to certain actions and situations and the end result of that calculation decides what the computer opponent will do. the "AI" will never decide what is more important or should be done first , it was already decided by the devs. so when we ask for better AI , what do we ask for? a real AI or someone decide better actions , to win the game and not try to make it accesible to a lot of people ( including children ) .

 

[jasper]

Singleplayer here. Complexity please.

Most of the fun comes from planning and developing my civ by taking all the elements or mechanics into consideration and trying ti connect the pieces and put a plan together.

Whats good about civ is i think they are in the sweet spot of complexity. Games like iron hearts 4 turn me off because its too complex. I dont want to tie my soldiers boots and make sure they have their broccoli but i do want some mechanics to fiddle around with.

Afaic, the ai is merely the supporting cast. They arnt actually going to win; i am. They only need to read their lines and put in a "good fight". Just find a good balance between letting the ai keep up with quick enough turns.

 

[Gedemon]

There is another alternative with asymmetric gameplay, where the AI doesn't follow the same rules as the human.

You can keep the game complex while still providing a challenge, Barbarians already work that way in Civ.

I'm sure better AI can be developed but using more modern techniques and machine-learning : take a holistic approach and log / analyze all the real games played (uploading anonymous detailed stats over time) and the AI will get smarter over time - possibly, unbeatable unless player gets very lucky.

I am sure this more modern approach would be better suited than the current if-then static set of rules.

Complexity is what makes the game interesting: please do not degrade it!

The game is too complex for today's machine learning, and if it wasn't there won't be a lot of player wanting to play against it. Note that it could be a good development tool to detect possible exploits early.

 

[Elhoim]

Complex game mechanics or good AI?

Yes.

There is another alternative with asymmetric gameplay, where the AI doesn't follow the same rules as the human.

You can keep the game complex while still providing a challenge, Barbarians already work that way in Civ.

That's what I prefer as well. One could say the AI currently works as "bots" in a PvP game, while an alternate development path to make the AI in single player more like PvE.

In any case, the devs intent seems to be to continue on the path of AI being other players, so yeah, more time spent in AI would be good, as it has been shown that an AI that can handle 1UPT combat is possible, and that's the most complex system.

 

[uhu]

the ai is merely the supporting cast. They arnt actually going to win; i am.They only need to read their lines and put in a "good fight".

!!

Civ 6 has still an issue with difficulty levels. New players have easier ones, which do not overwhelm them, good.

But really! There is something wrong, if eg. Marbozir plays on youtube at deity level. And then one of the animated 3D warmongers does his best to threaten him face to face. And then Marbozir just mumbles seriously bored: "Yeah, whatever you say ..." (featuring his typical small military)
I mean, who of us IS deity?! You really only need 4 difficulty levels - more do not hurt. The highest should be hard to beat even for veteran players.

edit: so yeah, more time spent in AI would be good, Unaltered Gameplay projects could bring about substantial improvement (again)

 

[PiR]

So what would you prefer? I don´t see how the devs could create an AI that handles all the additions to the game in a way which the playerbase would regards as competent. It would indeed be an amazing feat if they did but nothing worth hoping for this late in the game lifetime cycle.

Are you talking about Civ6 ?

 

[Linklite]

There is another alternative with asymmetric gameplay, where the AI doesn't follow the same rules as the human.

You can keep the game complex while still providing a challenge, Barbarians already work that way in Civ.

To a degree. In AoEII, the AI created units without resources, villagers, beyond building, were just for show. It was evident even to 10yo me because I would capture and hold all their resources and theydstill pump out units. I'd like to avoid that if possible - being able to deny enemies resources us half the attraction of warfare.

 

[Archon_Wing]

The Aoe2 AI only cheated on the highest difficulty, by giving themselves extra resources per age, but it wasn't infinite. It is possible to keep making stuff by selling at the market, even without gathering though.

Newer versions of the game (HD and Definitive Edition) don't cheat at all though can micro their units faster than a person would be able to.

Incidentally, the newer AI is harder than the older, cheating AI even without cheating.

 

[Gedemon]

To a degree. In AoEII, the AI created units without resources, villagers, beyond building, were just for show. It was evident even to 10yo me because I would capture and hold all their resources and theydstill pump out units. I'd like to avoid that if possible - being able to deny enemies resources us half the attraction of warfare.

Not that kind of rules, or not for all the AIs, as you said there are already Barbarians or City States in the game following different rules.

It's maybe more a variation of the first proposition in OP than an alternative, accepting that the AI can't win the game by reaching one of the game's state an human will target (ie a "victory type") but still be obstacle on the road for that target, being an entity (for example a "Nation", imagine a CS with multiple cities) that follows less complex rules, the more complex rules used by the humans being the tools to reach one of the victory.

 

[Alaindor]

The game is too complex for today's machine learning, and if it wasn't there won't be a lot of player wanting to play against it. Note that it could be a good development tool to detect possible exploits early.

I don't think so... it is amazing what one can achieve using machine learning techniques nowadays...

To support this hypothesis, this link tells us that AI machine learning could beat humans 80% of the time on Freeciv, in 2016: https://techcrunch.com/2016/12/06/a...-human-players-at-complex-civ-strategy-games/

Granted, Civ6 is more complex, but that was a whopping 5 years ago... eternity, as AI goes, and ML has vastly improved these last few years.

 

[Aristos]

Why should it be OR?

The very example you use, Vox Populi, made the game more complex AND the AI much better.

It's not a dilemma, it may be a matter of competency. Yeah, yeah, I know, the suits this the suits that, but when your AI does not even improve basic resources anymore (as has been widely reported), well, sorry, but not sorry for doubting your competency...

 

[dagriggstar]

I'd prefer a simplified initial release for civ 7. Both to help the AI and to help new players learn the game.

I also don't think a new mechanic should be introduced without at least one faction having some sort of bonus relating to it. If I asked you which civ was the best at harvesting well.......

 

[aieeegrunt]

Complicated does not mean better, and it certainly does not mean deep.

Civ6 is conplicated in the sense of being a minimanaging clickfest, but the gameplay is not deep at all

Expand as much as possible

Minimax yields

That is literally all there is to this game.

How complex are the rules for chess or Go?

 

[PiR]

Expand as much as possible

Minimax yields

That is literally all there is to this game.

How complex are the rules for chess or Go?

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