The AI

darko82

Emperor
Joined
Dec 5, 2005
Messages
1,328
Location
Poland
The AI is an offence to any fan of the series; if Firaxis changes the mechanics, and implements a different system; they should be sure how to make it work properly; they messed it up; and abandoned the game, basically. This game is huge and complicated; but the state, at which they released this game is not accaptable; people should not tolerate this practice; people should not buy the game at release and never pre-order because there is a big chance is it not working as supposed to. Furthermore, many important issues like the AI actually never get fixed. Modding can improve some things, but let's be fair; they are very limited (time, money etc.). Civilization has unlimited potential; a great game, no question, but the things and how they handle them is not acceptable. They should focus more on the game quality, not money. Releasing half-products only destroys the game and franchise reputation... In my opinion, Firaxis failed with Civ5.
 
I agree. Not because of the AI, but because of how poorly optimized the game is. It's why I have no interest right this second in "Civ 6" and discourage others from "wanting it" while knowing nothing about it other than its name. Every time I mention this, I invite people to tell me that with BE, they've offered a well-optimized, 64-bit game, which would indicate they've learned from their mistakes. Nobody has taken me up on that offer though, which doesn't instill a lot of confidence in me.
 
I agree. Not because of the AI, but because of how poorly optimized the game is. It's why I have no interest right this second in "Civ 6" and discourage others from "wanting it" while knowing nothing about it other than its name. Every time I mention this, I invite people to tell me that with BE, they've offered a well-optimized, 64-bit game, which would indicate they've learned from their mistakes. Nobody has taken me up on that offer though, which doesn't instill a lot of confidence in me.

It is disappointing that they didn't make a 64 bit version of CiV. However, I can't imagine they wouldn't make a 64 bit version of Civ 6. It does run quite fast now on my computer after all of the patches compared to the base when it launched. I'm still using the same computer that I had when it first came out without any upgrades, and there is a significant difference in turn time.

I also think it's unrealistic to think they'd redesign the base game engine for BE and change to a 64 bit game. I see BE as a cash cow that extends the life of the CiV engine.

For a better AI experience look at the Community Patch Project, it's revitalized the game for me. The AI is definitely better than the base game there.
 
I agree. Not because of the AI, but because of how poorly optimized the game is. It's why I have no interest right this second in "Civ 6" and discourage others from "wanting it" while knowing nothing about it other than its name. Every time I mention this, I invite people to tell me that with BE, they've offered a well-optimized, 64-bit game, which would indicate they've learned from their mistakes. Nobody has taken me up on that offer though, which doesn't instill a lot of confidence in me.

At first, I didn't see the big deal here, regarding 64-bit ciV. After a bit of research, I've realized that firaxis just dropped the ball here. There just isn't a good reason why they didn't do it, other than 'don wan to, don hav to'. My computer is outstanding, and the only game that runs like crap is ciV. This really sucks because it's the game I play the most by far. Having a five minute load screen at launch is ridiculous, because I can see the memory and cpu are IDLE!!!
 
At first, I didn't see the big deal here, regarding 64-bit ciV. After a bit of research, I've realized that firaxis just dropped the ball here. There just isn't a good reason why they didn't do it, other than 'don wan to, don hav to'. My computer is outstanding, and the only game that runs like crap is ciV. This really sucks because it's the game I play the most by far. Having a five minute load screen at launch is ridiculous, because I can see the memory and cpu are IDLE!!!

Yep. On top of that, my PC turns into a stove. I have some really good hardware, but play with the settings down just to save on energy. What a bizzarre thing to say about a psuedo-3d game.
 
Agreed. I had to chuckle at the thread awhile back where people listed favorite overheating stories. It's 60 degrees (F) where I live this evening...

... I am wearing a sweater but still have a dedicated floor fan blowing into the back of my PC. All because firaxis decided that 64-bit was just a fad.
 
Any examples?

Don't tell me you need examples of dumb, incompetent AI, which makes basic errors and mistakes in different respects. But of course, the tactical AI is what I am especially reffering to, because it is sooo weak.

CP is improving it in some ways, but still, if it wasn't for Gazebo, Firaxis would not move a finger on it. They just like that left it malfunctioning - after getting all that cash. That is the point. Paying Gazebo would be a better spent money than paying Firaxis for releasing a broken/unfinished game.
 
Don't tell me you need examples of dumb, incompetent AI, which makes basic errors and mistakes in different respects.

Actually, examples would be helpful. I'm always curious about what gameplay issues people have.

The tactical AI can be weak but it's immensely improved from what it was at release.

The one thing I'll say in support of Firaxis is they don't have unlimited time or money. To develop a flawless tactical AI would take so much time and money that we'd probably still be waiting for Civ V to be released if they decided to undertake this. In the meantime, they'd go bankrupt. Sometimes you have to make compromises.
And please don't tell me it's easy to program a tactical AI, if it was easy it would have been done.
 
The one thing I'll say in support of Firaxis is they don't have unlimited time or money. To develop a flawless tactical AI would take so much time and money that we'd probably still be waiting for Civ V to be released if they decided to undertake this. In the meantime, they'd go bankrupt. Sometimes you have to make compromises.
And please don't tell me it's easy to program a tactical AI, if it was easy it would have been done.

It has been done, which is where your theory fails. There are a number of AI mods available, all authored by people who did so with no expectation of ever seeing a dime for their efforts.
 
Firaxis is just hiding behind a excuse that it'll take so much time and money to make improvements to the military AI. They just don't wanna. It has been proven that military AI can be improved by modders.
 
Hey I know the AI is pretty bad at war, but...

1) I still get crushed occasionally by carpets of doom, which I think is an acceptable compromise (machine gets more units, human gets brain)

2) The work done by modders (bless you all!) is entirely predicated on the previous work done by Firaxis. I don't excuse sloppiness but let's give some credit for a really sweeping game that gets way more right than it does wrong.

(I am hoping this post qualifies me for free civ vi. I sound like such a brown noser.)
 
I don't really mind the AI not playing very well on release. I'm more sadden by the fact Firaxis patch the game once or twice per expansion and then forgets about it. It's a bit of a shame considering how much Civ5 is played.

Balance coupled with AI improvements will provide people with the singleplayer challenge they seek. Both are done through regular patching. Humans learn as they play to know what is good or bad. The AI doesn't have this advantage so it must be emulated through patching its behavior.
 
I don't know if this is common practice, but does firaxis offer financial compensation to modders in order to buy their work and fold it into a patch? I know some gaming studios will take the best ideas from their modding community and pay to use them in a patch or expansion. I have seen some impressive ai tweaks (acken's included) and it seems like firaxis is missing an opportunity to more frequently patch the game without dedicating programming hours that are going to (hopefully) a solid civ6.
 
It seems to me that Firaxis is very pragmatic in the sense that they invest minimal time for a high return.

The customer base has demonstrated, repeatedly, year-after-year, that we will pay full MSRP for beta quality. What possible motivation does a for-profit company with this kind of proven track record have for doing things differently?

They have no direct profit motive to issue balance patches. Frankly, I am quite grateful for the ones we have gotten. But my expectations are low. Still, the entertainment value from what I paid has been terrific.

I worry that Civ 6 will be a subscription model...

Firaxis is just hiding behind a excuse that it'll take so much time and money to make improvements to the military AI.

Citation please? When or where have they even bothered making such an excuse? Does the question even come up a press conferences?
 
If I look into Civ5 sources I expect to see smth like for military AI unit:
if (canAttack) {
attack();
} else {
randomMove();
}
And for governor:
build(getRandomBuilding());

Of course, it is not exactly like this, but very-very close to. Just after playing several games it becomes very clear how to fix at least half of AI stupid moves, but this wasn't done.

And what they did with this IMPROVE OF MP GAMING? When 99% of games ended in crash after 4-5 turns... Just I hope all this guys will be watching 'Configuring Game Data' for ages while burning in hell.
 
If I look into Civ5 sources I expect to see smth like for military AI unit:
if (canAttack) {
attack();
} else {
randomMove();
}

It's a hell lot more complicated than that.... think spaghetti... a very big bowl of spaghetti. I assume you have access to the Civ 5 SDK (CvGameCore in particular)?

You have at least 3 different AI classes handling unit movement (CvAIOperation, CvTacticalAI, CvHomelandAI, possibly some others that I don't remember off the top of my head. From what I remember off the top of my head, TacticalAI generally handles offensive operations, while HomelandAI generally handles the home-front as well as exploration).

Each type of operation is pre-assigned a bunch of targets and each target is pre-assigned some units at the beginning of the turn (in other words, if the AI captures a city, it will still feel "threatened" by that captured city until the end of the turn. Units that were "drafted" for the purpose of fighting that city won't be re-assigned until the next turn).

CvMilitaryAI is an "overseer" that computes a few not-so-useful statistics, and handles some general high level decisions (e.g. 'do we have enough units right now').

Each "type" of move (generally, they are stuff like: go heal, go plunder a trade route, go attack a city) is handled sequentially and independently of one another, with 'priority' of what type of move to take set in stone from turn 1 (through a combination of XML priority weights, and a bit of the AI flavors). Barbarian moves are handled sequentially based solely on their XML priority weights..... be careful if you ever mess with those, as it can lead to some very stupid barbarian maneuvers.

There is no random move, any units that "slip through" are assigned a skip-turn mission.
 
@Thorz, if it is not too much of an ask, can you describe how the AI ends up throwing units in the water? I wish I bookmarked that explanation when I last saw it!

My recollection is that this terribly counterproductive behavior derives from a chain of superficially logical algorithms. Talk about your spaghetti code!
 
@Thorz, if it is not too much of an ask, can you describe how the AI ends up throwing units in the water? I wish I bookmarked that explanation when I last saw it!

My recollection is that this terribly counterproductive behavior derives from a chain of superficially logical algorithms. Talk about your spaghetti code!

I'm pretty sure Delnar_Ersike gave an explanation for this somewhere, you can sift through his posts if you're that interested :p. My guess is that the pathfinder does not weight water tiles for land combat units properly while the units attempt to re-position around (or "move towards") their target, coupled with a dysfunctional DangerPlots system.
 
Top Bottom