Can we do anything to make Firaxis devote more resources to improving AI?

[Japper007]

There are some obvious AI issues that need to be tackled. While I don't expect Firaxis to ever get the combat AI working properly, some things should be quite an easy fix, such as:

-AI expands like it's playing high-level civ 5 Brave New World (which is to say very little). I shouldn't be able to crowd out several Diety AI's on a small Pangea, but I can routinely pull it off.
-AI over-prioritises Religion and undervalues the other districts (just look at Kongo, he stands out as a good AI because he cant build Holy Sites)

There is tons of gain to be made by just fiddling with the priorities/values the AI assign, neither of these things need any new code, just some tweaking the numbers.

-AI doesn't escort settlers and builders properly

Now this is slightly harder to fix, but it was eventually managed in civ 5 too. It is also one of the biggest immersion breakers in the game currently. It is guaranteed to get picked up on by even the lowest skill-level players within the first few games.

I'd also actually be fine with logging game-stats for raw data, as long as it doesn't make the game even slower than it is already and it's optional. Man of War uses a similar system to great effect, it has one of the "smartest" AI's in the business.

I think the biggest issue however, is Firaxis lack of communication with the fanbase and slow patching regime. This isn't the 2000's when you could get away with a patch a year! Firaxis is competing now with the likes of Paradox and Amplitude/Games2Gether, who actually get into the forums and implement fan-suggestions or even incorporate fan-made mods on a regular basis. They need to step their game up!

 

[Krajzen]

I think the biggest issue however, is Firaxis lack of communication with the fanbase and slow patching regime. This isn't the 2000's when you could get away with a patch a year! Firaxis is competing now with the likes of Paradox and Amplitude/Games2Gether, who actually get into the forums and implement fan-suggestions or even incorporate fan-made mods on a regular basis. They need to step their game up!

Yeah. I am frequent visitor on Paradox forums and the difference between them and Firaxis is enormous.
Paradox is a gaming company of 21st century, constantly developing their games with regular DLCs and free patches while informing fanbase about every upcoming new system with dev diaries and faq.

Meanwhile Firaxis (or 2k games) is stuck in the 20th century - silence on all channels and then sudden drop of a patch or a DLC, then back to the shadows, nonexistence on forums, no dev diaries at all, rare patches, archaic model of two box expansions per game cycle...

Being completely silent about very pld complaints and requests.
Artificial Intelligence. Diplomacy. Hall of Fame. In game stats (like Info Addict). Canals. User Interface. Modding support (dll code). Map seed. Jesus Christ.

 

[Kouvb593kdnuewnd]

Yeah. I am frequent visitor on Paradox forums and the difference between them and Firaxis is enormous.
Paradox is a gaming company of 21st century, constantly developing their games with regular DLCs and free patches while informing fanbase about every upcoming new system with dev diaries and faq.

Meanwhile Firaxis (or 2k games) is stuck in the 20th century - silence on all channels and then sudden drop of a patch or a DLC, then back to the shadows, nonexistence on forums, no dev diaries at all, rare patches, archaic model of two box expansions per game cycle...

Being completely silent about very pld complaints and requests.
Artificial Intelligence. Diplomacy. Hall of Fame. In game stats (like Info Addict). Canals. User Interface. Modding support (dll code). Map seed. Jesus Christ.

Just compare the new Stellaris DLC against Rise and Fall and the difference seems to be huge. Paradox is willing to change stuff they don't feel work well and really change the game while Rise and Fall just feels like it is going to be an addition which is not in any way going to fix the core issue such as chopping. Also paradox explain very well why they are doing all these changes while you have no idea what Rise and Fall is really about.

So that is an area they could really improve in.

 

[heinous_hat]

While I completely agree with most criticism regarding the poor performance of various AI systems, there's one important thing to not lose sight of...

Firaxis releases source code. Few other developers do this, and do it with regularity. We're quite fortunate to have this made available, even if the release lags behind that of the game.

 

[IntoTheSky]

My post from some other thread.
Should be posted here.
Hello.

Is this discussion going in the right direction? I don't want to say it is not interesting, etc. but I can't see many voices demanding FIXING what we already have.

Bugfixes.
This is priority. Without fixes & changes you will have your new Civs and new cool leaders that are not even improving terrain around their cities ( on Deighty)
I don't need new civs, leaders and exciting features now.
I can live without Hwacha and Seowon.
I want IMPROVEMENT,

My wishlist.
1. AI and units.
From workers and scouts to modern artillery - I want to see AI using his units smart.

What concerns me , are live streams from Firaxis.
I watched it . 2 guys from Firaxis were excited about Hwachas and Seowons . Meanwhile, they got war with Egipt and Greece.
Egipt started with loosing 2 chariots in very stupid way. They died from city barrage + archer.
Then greek catapult arrived. AI put this catapult in "death hex" , loosing it instantly. Then he lost 3 more units in 3 turns.
Civ VI ( Deighty) is like playing with not very smart 10-years old kid.


Have you heard what MAIN DEVELOPER of this game said during second stream ? ( around 28min10sec of stream)

Ed : "Ok , so we have, uhm, the inspiration for Feudalism and that means that as soon as we can finish that civic , we will go ahead and get the stirrup technology, so we can move on and work on some other tech instead".
Main Dev : " This is why Ed can play on higher difficulties , where ...( laughing) I'm not." !!!!
Ed: " Eee..a lot of people have learned this way to play. It's nothing too special "

WTH , I say?
This guy is main developer , and he can't understand some very basics of this game.

I was hoping for some serious people, but I saw some bald guy excited about "new exciting features" and Hwachas.

I was hoping that developers treat us serious, but they are not.
Im affraid that in Rise and Fall I will again slaughter brainless AI units with my smaller , undertech army.
But Moderator Action: <snip> it! I will have Hwacha and this excting Seowon!

In some games, Deighty AI is in constant troubles with barbs.
Barbs are as stupid as leaders, so they are slaughtering each other in some nonsence battles.
I would say its good that barbs can;t capture cities ( which, ofc. isn't good, its stupid) , becouse otherwise they would own AI players.
How do I make my "Deighty" games a little bit harder ?
I turn off barbs.


Workers.
Yes, we have workers in this game.
I saw "Deighty" AI in information era with his cities in the middle of jungles , only with some districts&wonders.
No mines, no farms, no Moderator Action: <snip> improvements. What is this, protest against "Global warming " or a game of Sid Meier Civilization?
Sid, this is under your name. Do smth, man. I know you can, you are a great person

2. AI and war.
This is still point 1, in fact.

Do you know, how I make "Deighty" a little bit harder?
I'm not building any Encampent districts and any Walls.
Now, Im considering avoiding this cultural tech which gives cities ability to defend.

Two well-placed Encampments can hold 10 AI "Deighty" players for a whole game.
AI fights Encampments to his lost drop of blood. He can sacrifice his whole army, and not even destroy your fort. Pathetic.
Artillery is not for AI. Too complcated. He just uses it as cannon fodder, his arties drive around without any sence.

Air units?
THERE IS NO AIR COMBAT IN THIS GAME.

Overall tactics?
Yes, we have tactics in Civ VI. Sean Bean is talking a lot about it.

3. Religion.
When I saw this religion units casting some spells I thought " Really? " , and I never build ANY religion district in my games.
Im not using religion at all. "Deighty" AI using religion a lot. This hordes of spell-casting monks are scarry.

WHAT IS THIS?
Sid.
I can't believe this Moderator Action: <snip> is under your name.
Will there be dragons, Leviahatannah and Behetomoth in CIV 7?

Ok, enough for now. Its just a peak of mountain

Moderator Action: Please review and comply with our rules regarding use of inappropriate language. Browd
Please read the forum rules: http://forums.civfanatics.com/showthread.php?t=422889

 

[kingchris20]

Probably the best way is to list all the problems with the AI that could be improved upon as a community. Then find out what parts of the coding is causing it, as a community. Then create a detailed and properly coded fix, then send it to firaxis and ask them to check it out, make tweaks as they see fit and implement it into an update.

EDIT: ie fix it ourselves. Then if they were to implement it, they could make all the changes dependent upon selecting a new difficulty level above deity called (CivFanatic).

Firaxis time commitment = small to medium (quality assurance and big checking)
Firaxis financial impact = small (testing, approving, and delivery)
Firaxis rep gained = major (Civ fans will be once again 100% happy with the game)

 

[Art Morte]

The AI is the reason I won't be buying Civ7 when it comes out, whenever that may be. Two consecutive games with awful AI is enough for me.

 

[Ferocitus]

So it can be done, why Firaxis is not doing it?

Premature optimization is a sin. - Donald Knuth.

Because customers allow it.

Some customers want a new version as soon as they can get it, and
are prepared to accept the many inevitable flaws and inadequacies.
Some people think they have the right to deny them that considered
choice.

 

[hollus]

My fix (not going to happen): Firaxis should make it mandatory for anyone working on AI behavior and game balance to spend 2 working hours everyday playing the game.

Then there’s the original sin here: everything in this game is designed as if one would want to go tall. The AI goes tall. Districts, housing... tall-hungry.... but the winning strategy: wide! I should know, I played tall my first games, but then checked here and well..,

Now, a fix hoping they, indeed, read here. I am in the camp that believes that war weariness is behind many AI weaknesses like the units not taking the free shot (it is not free, it pays some weariness). But the biggest fixable sin: the mechanic where losing a battle incurs a lot more weariness than winning it.
The human player tends to be in the winning side. Say you play with 10 civs... the human is likely to be in the top 3, as it is no fun to be an underdog in such a long game and one can simply level down. As for the AIs, most of them are often losing, and, by definition, half of them lose badly, big and quickly.., and get huge war weariness in the peocess.
Why differentiate the two? It is realistic regarding real life, yes, but how many people notice? how many know the detais?
I suspect that just making war weariness flat with the number of combats (even better, turns at war), would likely go a lot way to make sure that when the AI is waging war, it is actually trying to do so boldly and effectively.

 

[Spudsie74]

It would be nice if the developers spent a little more time playing the game, but that's just not the way the world works. It's about maximizing profits, corporate mobility (moving up the Firaxis food chain), along with other annoying real-world issues which put athe customer about where the average American citizen is WRT to how their country is run (our opinion counts for little). Like the US, Civ 6 is NOT a functioning democracy, but an (Oligarchy? Despotism?). Take your pick...the development of this game is IMO on autopilot, in that what matters now (to the bean-counters) is the second expansion, and to a lesser extent, more DLC's.

Isn't mass-marketing wonderful?

On the bright side, like the percentage of citizens that actually vote, SOME folks actually care about the consumer, and there is a (jet-lagged) response to customer feedback.

It's a bit like expecting Star Wars VI to be as good as the first one. Nominations are open for the next Civ 6 leader...Jar-Jar Binks.

 

[Ever Adrift]

It would be nice if the developers spent a little more time playing the game, but that's just not the way the world works. It's about maximizing profits, corporate mobility (moving up the Firaxis food chain), along with other annoying real-world issues which put athe customer about where the average American citizen is WRT to how their country is run (our opinion counts for little). Like the US, Civ 6 is NOT a functioning democracy, but an (Oligarchy? Despotism?). Take your pick...the development of this game is IMO on autopilot, in that what matters now (to the bean-counters) is the second expansion, and to a lesser extent, more DLC's.

Isn't mass-marketing wonderful?

On the bright side, like the percentage of citizens that actually vote, SOME folks actually care about the consumer, and there is a (jet-lagged) response to customer feedback.

It's a bit like expecting Star Wars VI to be as good as the first one. Nominations are open for the next Civ 6 leader...Jar-Jar Binks.

But this isn't inevitable, even if it is the case with Civ VI. Firaxis doesn't seem to care and has incredibly poor customer service - the fact that they don't hotfix even when there are game-breaking problems (e.g. civ placement on random maps if not using YnAMP) demonstrates this. Part of it is running on the franchise - Civilization used to get a lot more love and care, but the franchise is well-established and well-loved now and many people buy it out of loyalty. The mod community helps fix problems and keeps players around for the expansions.

I really wish Firaxis did a better job. There are a lot of problems that shouldn't take much work to fix - including the AI. When you look at online games like Warcraft and FF14 - their gaming communities are furious if it takes more than a day or two to hotfix even minor problems. Indeed, hotfixes occur at least weekly and the developers clearly read gamer feedback and do their best to act on it.

I don't understand why Firaxis can't dedicate a few resources to maintenance and fan/developer communication...

 

[The Technomancer]

The issue really comes down to it being really, really hard to build a good AI as compared to a perfect AI.

Throw the game at a machine learning model long enough with enough compute power behind it, and you'll have a perfect AI that will be nearly impossible to beat.

I can expand on the idea if people would like, but building artificial stupidity (as in, ensuring the AI is fallible and beatable by a human without being too stupid) is a hell of a lot harder than building artificial intelligence nowadays.

 

[Alexander's Hetaroi]

I was going to post this in another thread, but then it closed as I was making it. This is a way that the AI could possibly be more competent going for a certain type of victory.
It could be like religion that was updated in the fall patch where there are about 9 leaders who will not go for one unless certain things happen: faith based hidden agenda, find relic, natural wonder (this might not be a eureka anymore for astrology?), any source of early faith.
Domination as a major choice would be best suited for Macedon, Rome, Aztec, Persia, Mongolia, Norway, Sumeria, Scythia and Chandragupta.
Indonesia for Island Plates, England (if more than half of the civs are on a different continent than her), Spain (if it suits his religious game) and Germany (if the world becomes friends with city-states. (Well that ended up being a lot )
For Culture: if you build any type of wonder early on or, find an early source of culture other than city center, like natural wonder or luxury, or a relic as well.
Science is harder, but maybe for any civ that rushes to writing early on.
And of course different agendas can play into that as well. ​

 

[omega4]

Reposting this since my thread was locked for speaking the truth to power.

After 11 hours and 2100 turns, the year is 2847 AD. And no one is fighting.

I've forced each Civ to declare war on each other. Yet after a few turns, they all make peace with each other.

Rome has the most territory and cities, yet refuses to destroy the smaller, weaker civs around it.

Looks like I'll be headed back to Civ5 Vox Populi.

 

[omega4]

While I completely agree with most criticism regarding the poor performance of various AI systems, there's one important thing to not lose sight of...

Firaxis releases source code. Few other developers do this, and do it with regularity. We're quite fortunate to have this made available, even if the release lags behind that of the game.

So what you're saying is Firaxis puts out a half finished game knowing that modders will complete the job for free?

 

[Krajzen]

My biggest fear for Rise and Fall is simple, new features won't matter because of useless AI. I seriously think fanbase here and on reddit should made big uproar at some point of horrid AI... Devs are present here enough to get Tamar of Georgia meme, well maybe at some point they will get "improve the AI" meme.

By the way, when did firaxis release source code of civ5, after bnw?

 

[pietro1990]

My biggest fear for Rise and Fall is simple, new features won't matter because of useless AI. I seriously think fanbase here and on reddit should made big uproar at some point of horrid AI... Devs are present here enough to get Tamar of Georgia meme, well maybe at some point they will get "improve the AI" meme.

By the way, when did firaxis release source code of civ5, after bnw?

I think the new alliance's system will be useless because of the bad diplomacy. for example the warmonger penalties that aren't fixed.



In current game its impossible to keep a alliance the entire game because once you start conquering cities they don’t want be you're ally again.. I loved how you could win the game in civ 4 by forming teams inside the game ask a alliance if the AI was really positive about you...

Even if you don't warmonger and play peaceful Ai will sometimes don't want to be you’re ally anymore because you don't comply with their agenda anymore ( for example they like you for having lots of military then when you don't built mroe units they start denouncing you.

Oh yeah lets not forget that the only way the AI declares war is through formal war in the trading deal wich results in even friendly Ai declare war on you just to get a trade deal.. I’ve had Indonesia in my game declare a formal war 5 TIMES and afterwards wanting to be friends each time really?

i've had AI who only had horseman as armies declare war on me with a other AI while i had infantry because hey i can declare war with lower warmonger penalty why not..

They need to change this whole causic bellic, warmonger penalty and agenda system because it making the AI act like a crazy person.

 

[playshogi]

The best chance for great AI is for Google to buy the game at let loose Alpha Zero on it for a few days.

 

[Cronos]

I fired up a game to see if I could get back into civ ahead of the expansion.

The ai does some things well. It can build cities and infrastructure. Im testing it by trying to win without building a single unit. I have been subject to multiple DOWs but the ai is not able to take any of my cities. There is literally no threat from the ai. So none of my decisions matter at all i literally cant lose.

Further i get random messages i have been denounced for having no military and the next turn denounced for being a threat! And then same ai commends me the next turn for having a great research rate. Unfortunately the diplomatic and militaristic systems are completely broken for me which should be fundamentels of civ

Edit: pericles actaully attacked my capital! Go for it my ancient friend. But alas no with defences over half down next turn he sued for peace and gave me everything. Potatoes

 

[glider1]

The best chance for great AI is for Google to buy the game at let loose Alpha Zero on it for a few days.

What will happen is game API will be written so that Alpha Zero (or equivalent) neural net can self learn the game for x-hours for each difficulty level. Firaxis will contract Google (or equivalent) for the emergent neural network code and plug it into the game. Firaxis will not write any AI code and will concentrate on game mechanics and interface code. If nothing disruptive happens, we could see that happening for civ 7 already. It may take civ 8 to happen since there will be a lot of demand for Alpha-Zero mainframes from the business/governance/military sectors that will mean games might have to take a back seat.

But yes, there is no point hand-coding decent AI's any more for the really big gaming companies. It is costly, complex, time consuming and bug ridden. The quality of AI from the neural network mainframes will probably be a little better than the best manually coded AI's ever written (like VP) but not that much better either since it is not worth Firaxis to pay for too much money for computing time on the neural net mainframe.