[GS] Domination, air combat, AI and the new era

[Cagarustus]

They haven't even mentioned AI in the new releases, that I've seen.

They said that they would improve ‘existing game systems.’ One could include AI into that. The issue is that Fireaxis wouldn’t want to open up Pandora’s box by explicitly stating they’ve made AI improvements. I think paradox interactive made this mistake when they did it for HOI4 with regard to performance improvements.

 

[glider1]

Problem for me is that unless the AI is reasonable across land/air/sea, pushing toy soldiers around for the sake of it doesn't do it for me. Strategy games is a way to entertain and stimulate brain activity by being challenged to think dynamically as situations change according to the pressure of your opponent. Without an opponent, a strategy game is just a puzzle game.

But if they haven't concretely said they are going to work on the AI how can I be sure that they will do anything but a token gesture next expansion? If they haven't said they are going to release the code to enthusiasts to improve the AI how can I be sure they ever will?

So once again, there is no point investing in civ6 for people like me even when it is on sale. Just go play chess and VP. Simple.

 

[Roxlimn]

It depends on what you're after. Every Civ iteration, Civ 3 to Civ 5 for those who still have their memories here, ended with AI that were bad in the final expansion release. The Civ3 one remains terrible to this day. Once you can form an Army, the AI can't handle it and they auto-lose the domination game. You can take over at will. Civ4 and Civ5 were bad at the final release. You will still get recommendations today to download community content to improve the AI. Which still doesn't make them brilliant - just not terrible.

If past behavior is any indication, they can be expected to release modding capability some time after Gathering Storm, if it's the final version, and for the community patch to finalize the game some years after that, if there's any community interest in Civ6 after Gathering Storm. Civ 5 Vox Populi mods are still being updated as of early this year, so it's not quite final-final, and that gives you a broad timeline of what to expect and when.

Basically, Firaxis has always left it to the community to figure out the AI and there's every indication it will be the same this time around. While this does raise questions of where developer responsibility ends, there can be no question that the final AI mods are pretty savvy and usually include all the meta-plays that people who play the game obsessively can know.

 

[Trav'ling Canuck]

Problem is that Firaxis never get around to it because they are always adding features to increase sales and you cannot improve the AI properly until it is feature complete.

You can't hand-craft an AI algorithm for an incomplete ruleset. The rules have to be complete and unchangeable before you can make anything like a credible half-decent AI of that nature. This means it is necessarily work done after the last expansion is released

This is the core dilemma, the need for a final ruleset to truly optimize AI decisions, which runs at odds with the desire to throw more and more systems at the game every expansion.

I will say, though, that part of the issue isn't just not putting more resources into the AI during development. It's also not involving top quality players in the development of new expansions.

Having some better civ players involved would at least put someone in place who'd be in a better position than the developers themselves to recognize what the most efficient decisions are likely to be based on the course of development, and feed those ideas into the AI guy.

 

[Roxlimn]

Didn't they do that with BTS? Trovato was involved with Civ4 prior to the release of BTS and the AI was still bad. This is a development issue, but also a manpower issue. Blizzard devotes years, involves pro gamers, and lots of manpower to develop balanced MP content, and the AI for SC2 is STILL quite bad. The recent uptick in AI power in VP was because the modders corrected a build order bug in the AI, causing it to develop its cities suboptimally, which I suspect was only uncovered after the last patch was played for several hundred hours.

 

[ashendashin]

The recent uptick in AI power in VP was because the modders corrected a build order bug in the AI, causing it to develop its cities suboptimally, which I suspect was only uncovered after the last patch was played for several hundred hours.

Uh, what? I don't remember an individual uptick like that. VP is still getting updated every few weeks with AI improvements, and individual bugs hampering the AI due to legacy code are usually from untouched game mechanics as far as I know. I mean, VP doesn't even use the vanilla flavor system as the sole source of AI decisions, so I've no idea where you got the build order thing from.

 

[acluewithout]

Didn’t the guy who wasn’t Ed or Sarah in the let’s play mention buffs to the AI? I think he said they’re always working on stuff including the AI?

This is the core dilemma, the need for a final ruleset to truly optimize AI decisions, which runs at odds with the desire to throw more and more systems at the game every expansion.

I will say, though, that part of the issue isn't just not putting more resources into the AI during development. It's also not involving top quality players in the development of new expansions.

Having some better civ players involved would at least put someone in place who'd be in a better position than the developers themselves to recognize what the most efficient decisions are likely to be based on the course of development, and feed those ideas into the AI guy.

Yup. Hard to program the AI when the rules keep changing. Victoria also mentioned problems with AI and planes being maybe about the AI turns timing out - that could be a thing.

I really don’t think the AI is that bad. But moreover, I can’t see how it could be made all that could economically unless that work is done by the community. The meta for the game keeps changing even without the rules changing. I can’t see how a games publisher could keep up with that.

 

[Tech Osen]

I'm no expert but it's hard for me to imagine that it would be difficult to make the AI attack with ranged units before melee units. It's the very basics that still don't go properly.
The other day Shaka came with a huge armada and one mech infantry. My city almost fell and then he attacked with the mech first, killing it without taking the city and I could take out the four nucleair submarine armadas at my leisure.

 

[V. Soma]

I'm no expert but it's hard for me to imagine that it would be difficult to make the AI attack with ranged units before melee units. It's the very basics that still don't go properly.
The other day Shaka came with a huge armada and one mech infantry. My city almost fell and then he attacked with the mech first, killing it without taking the city and I could take out the four nucleair submarine armadas at my leisure.

This

 

[Victoria]

I do feel as if the AI being bad is a good thing. Certainly as I play devils advocate in my head it makes sense. It is all about cool new toys and saleability. When people make a mistake and the AI should punish them but does not, it’s that naughty happy feeling

 

[craney1987]

Would a competent AI not be beyond a casual players PCs capabilities?
I understand the need for improvement here. I should not be able to beat deity. Yet I do.
But I also think that I struggle to run the game now, let alone with all the extra processing time a competent AI would need.

 

[glider1]

Vox Populi showed that you can make a significantly better AI that runs faster than vanilla with the same specs as vanilla.

 

[craney1987]

Then firaxis can do better. I assume they have a few more resources up their sleeve.

Surely.

 

[Xur]

The AI should process strategy/needs/wants 24/7, also during player turns. The AI should put units on sentry/fortify instead of moving them mindlessly around constantly. The early AI should have some sort of idea of who its neighbours are and not to start wars 60 tiles away, that it cannot possibly end. The late AI should build up for wars, start that war and win that war. It feels like playing against a monkey that just happens to have 80% build bonus and therefore it almost cant NOT have an army. This monkey then rolls his head all over the keyboard and hopes he wins - It's not an AI with a plan. It might as well just be a RNG building stuff and sending it in randoms directions. Any improvement on the AI is good (or required).

If the game had less gameplay features and a better AI, it would be a better game. AI is that important.

 

[Victoria]

I assume they have a few more resources up their sleeve.

Surely.

Well they have many artists and play testers

 

[Stilgar08]

I do feel as if the AI being bad is a good thing. Certainly as I play devils advocate in my head it makes sense. It is all about cool new toys and saleability. When people make a mistake and the AI should punish them but does not, it’s that naughty happy feeling

To me that's not working. I feel like I'm cheating because the game is too easy, dark ages aren't dark enough and AI is just there to get bullied around at my command...

 

[glider1]

The AI should process strategy/needs/wants 24/7, also during player turns. The AI should put units on sentry/fortify instead of moving them mindlessly around constantly. The early AI should have some sort of idea of who its neighbours are and not to start wars 60 tiles away, that it cannot possibly end. The late AI should build up for wars, start that war and win that war. It feels like playing against a monkey that just happens to have 80% build bonus and therefore it almost cant NOT have an army. This monkey then rolls his head all over the keyboard and hopes he wins - It's not an AI with a plan. It might as well just be a RNG building stuff and sending it in randoms directions. Any improvement on the AI is good (or required).

If the game had less gameplay features and a better AI, it would be a better game. AI is that important.

A lot of those fixes don't even have to need much thought. They simply have to download the Vox Populi code and copy the algorithms into the civ 6 code. VP solved all those issues a long time ago.

But I'm sure that they are conservative and worry that changing the AI could have unanticipated side effect including CTD, bugs and unhappy players that no longer feel that naughty happy feeling.

 

[Gedemon]

Then firaxis can do better. I assume they have a few more resources up their sleeve.

Surely.

They lack the most important one: time.

 

[Xmonger]

It's easy to be an armchair engineer. OK folks let's stop handwaving and get out the slide rules to discuss two approaches to AI - expert system and a DNN/Deep Neural Net/Machine Learning/ML.

DNN/ML
How: Connectionist solutions work by using massive amounts of data - the more the better. The network continues to improve with data and deeper layers (generally). This data would have to be tens/hundreds of thousands of human games, and your population spread would need to be sufficient (e.g. games in a variety of situations). OK fine, say somehow they've got a data set. And train it - how so? Start with the input and output layers of the network, the output is "win/lose" (2 neuron) and the input would be possible game states. What is a game state, well, let's take a similar approach to AlphaGO. where we have an input node for each tile on the board.

How many tiles in a normal sized map, anybody know? Go has 19x19, according to SX it's 80x52 or 4160. But we have to accommodate all cases including Huge games, which are 128x80, which is 10240 input nodes. Compare this to 361 for Go folks. Further, consider the states of the input nodes, for Go it's beautiful, just a binary 1/0 (black/white). A tile state in Civ consists of (which civ * which unit * which terrain * which improvement). AT LEAST, I'm surely missing some. So now we have, let's set, say 20 civs, 10 units, 10 improvements, I don't know maybe 10,000 combinations (just pulling this out of the air, somebody here could work it out). OK, so 10k * 10k is 1 million input nodes.

Good luck with that folks. Someday we'll have networks like this, not today.

Expert System
According to Wikipedia "The most common disadvantage cited for expert systems in the academic literature is the knowledge acquisition problem. Obtaining the time of domain experts for any software application is always difficult, but for expert systems it was especially difficult because the experts were by definition highly valued and in constant demand by the organization.". So, for improved AI they'd have to put their top people (e.g. Karl/Carl, Ed, etc) on building the database. I'd estimate this would be a many man year effort. And then because it has to scale with difficulty you have to have that built in somehow, meaning also have a database scaled choices (bad - good). And while you're doing all this it would consume so many resources I personally wouldn't add other features to the release, thereby pissing off the majority of your customers in favor of the handful of those who want a smart AI. Finally it would be fragile, the history of expert systems is littered with many examples.

Enough said and FWIW.

 

[FearSunn]

Oh common, another "expert" counting nodes. Decent 4x strategy game AI takes 7 MB (one mp3 track). Example: "Pandora: First Contact" custom made AI by one man nickname Ail.

PS: "Why new AIs are so powerful?" this is what people discussing in Steam Discussions about Pandora game. Could you imagine such thread title here on Civ6 forum?