WTH is so special about the AI?

[Dis]

I'm so tired of people say comapanies shouldn't release a game in this state. Well I agree. But complaining will accomplish nothing.

Boycott Firaxis people!!!!!! b!tching about Firaxis yet buying their game is just plain stupid. How do you like that? I called you stupid for buying a game. Haven't you people learned from SMAC. SMAC was the same exact thing. The game was unfinished and required several major patches. And that did not correct the problem.

If you want to buy a game that is finished when it hits stores, do not buy a Firaxis game. The solution is so simple. I can't figure out why no one else can figure it out. If no one buys their games, they might have the incentive to fix it.

So I no longer give a sh!t about how much money you had to spend on a game. It serves you right for buying the game opening week before any reviews are out. Maybe next time you will get a clue.

You all deserve a buggy game, because you are gullible enough to buy a new game the first 2 weeks of release.

[kailhun]

Originally posted by benjy
oh, and this happened on the opposite end of the continent to the roman border. i like to see the AI aggressvely pushing moving its borders, colonising open land, but sending a ship around the coast to bypass 25 Greek cities and building a city a 1000 miles from its nearest town is stupid.... i really hope they tone that down a bit.

Actually, building colonies like that was quite normal say 2000-3000 years ago. If you look at a map of the Meditaranian (sp?) of that period you will see empires which consists of a lot of single cities spread around the coast of the med. Sometimes a greek, roman, carthaginian, syrian etc. city would be build away from the main empire with another empire in between. But as no-one really controlled the land in between in didn't really matter. The age when empires actually control all the land that they coloured pink, red or green on their maps is fairly recent. I doubt most of the 19th/20th European empires did that when it came to their overseas territories.

Robert

[kailhun]

Originally posted by Dissident
everyone on this thread is wrong. felt better saying that

funny how you guys say civ2 ai was better. When was the last time you played civ2? Civ2 Ai blew donkey balls. Civ3 isn't much better. All of those are valid points. But the ai is slightly improved. It needs a lot of work though .

Lucky donkey.

Speaking of which:

Toby the donkey is standing in standing in a field, when a spaceship lands nearby. An alien comes out and walks over to Toby. 'Listen, Toby. I've got a joke that'll blow your balls off.', he says.
'I've already heard that one.', Toby replies.

Robert

[Dis]

now I got that off my chest

I just don't think it is possible to build a good ai, and include the depth of a civilization game.

[Dev]

Originally posted by Green Giant


I think someone tried to sue Origin a few years ago for the incredibly buggy release of UO. They lost.

They settled.

/dev

[Skanky Burns]

Civ 3 AI actually uses boats to invade other landmasses.
They protect their troop ships with military ships.
They bombard improvements around the cities to cause your empire damage.
They can launch well-organised strikes against your weakest city.

Yes, the AI isnt perfect. It isnt even great once you get to know it. However, it wipes the floor with the Civ 2 AI, which is about the best you can ask for.

Until someone writes an AI that learns and adapts, we just wont have an AI that can stand up to a human player. No point knocking this (very good) attempt by Soren.

Just saying this again, the Civ 3 AI wipes the floor with the Civ 2 AI.

[Fidel]

amen to that

[Moraelin]

I never said that it's WORSE than the Civ 2 AI, since OK, the Civ 2 AI was atrocious. Its only saving grace was that it cheated like a little piggy. Half the rules didn't even apply to the AI at all, and the other half were bent according to the difficulty level.

I'm just not sure it's much of an improvement in Civ 3. The rules are still bent to give a dumb AI a survival chance. In fact, my own observations seem to indicate that now even combat has the difficulty setting factored in. That would explain why some people defeat tanks with archers regularly, and others have their tanks routinely defeated by archers.

In all fairness, it seems to be slightly more fair this time, since at least the research rules now work both ways. They also give the player a small boost if he's fallen far behind the AI. And generally I don't mind the idea that it's cheaper to research something that's already been discovered by everyone else. In fact, I actually like it. (I don't like the hardcoded 32 and 4 turn caps, though, or any other hardcoded caps.)

But again, it looks to me like it wasn't so much of an actual improvement to the AI, as some clever tweaking at how the rules are bent.

[benjy]

kailhun - i take youre point. if thats true, i think this is one of the times id like to see fun gameplay win out over realism. it just isnt fun to have a land mass with 25 cities on it, then for some civ to come along and build a city on a tiny piece of land in the artificial area between 2 cultural borders. it forces you to make war with them, because even if they defect, its placement will impeed your city growth.

plus, i just dont think this is the way the border thing is intended to work.

'course i dont wanna be a whiner and ill live with it.... cant wait for the patch!!

[Kolyana]

Actually, building colonies like that was quite normal say 2000-3000 years ago.

Yeah, if there was a reason for it. You can bet your sweet eggs that Rome didn't ship troops, materials and people half way around the world ... just because a lump of grassland was going up for grabs. Resources drove this sort of colonization, and from that point of view and in that context I have no problems with the AI's odd behavior.

*HOWEVER*, the AI is obviously driven by different rules ... and it's so obvious that it detracts from the game.

If the AI went all out to build cities next to strategic resources, we'd be congratulating it on a job well done.

However, what we're talking about here is the AI building cities in the most useless places on the map, just because it's spare land.

I've been in the situation when I have dominated and grabbed and entire southern continent ... when years into the game the AI nations started sailing past south. Their destinations? Small 1 square unclaimed land masses on the coastlines and southern tips.

One claimed a tundra square with mountains all around it, another claimed the same 2 squares over ... so these two were now perched in the most stupid manner on the brink of my kingdom.

It doesn't matter that x-amount of years later I managed to drag them into my kingdom through culture, if anything this is yet another reason why the AI shouldn't do it.

I ended up parking troops on each square at the tips of my land ... not vast swarthes of land, mind you, just 1 square islands on the tips of the continent.

[Antonin]

Hey all you whiners!

Ever heard of E-bay? If you hate the game so much, sell it and move on.

[Barchan]

Hmm. I think this AI land grab "problem" really helps illustrate the limitations of artificial intelligence. Here, as I see it, is the problem in a nutshell:

The AI is programmed to expand. Some say at all costs, and that's as may be, but it's still driven to expand. The drive to expand, in and of itself, is not unreasonable. Frankly, it's essential to success. Strategically speaking, if you don' t take the land someone else will.

The AI can best be described as "hopelessly optimistic" in its view of the success of these niche cities. Sure, the terrain may suck. It may be surrounded by other Civs. It may in fact have no reasonable expectation of long-term survival. But for the price of a settler and a spearman, the AI is making a bet that there will be coal or oil or something good under the soil, or that your civilization will crumble and its city will be the oldest, most cultured city left standing (or only one left standing!). You may be laughing at this, but people waste good money on impossible odds every day. Have you ever bought a lottery ticket or put a quarter into a slot machine?

Due to this “hopeless optimism”, the AI cannot fully calculate the actual inevitability of the failure of its little niche towns. It has to assume that if there’s land, there’s a chance for success. I can’t imagine the algorithms the AI would have to run to make the same decisions a human player does regarding the placement of cities. We calculate many different factors: terrain, proximity to own/enemy cities, current relations, future expansion plans, emergency changes to plans based on current events and etc. I think it’s all beyond what we can ask computers to do at this time. Sure, we could trim the expansion back, but then you’d have the opposite problem of an AI that expands too slowly or methodically. For example, if the AI were forced to expand in a tethered manner (i.e. no more than five spaces between cities, the human player could race ahead, plant a few cities and cut the AI expansion off at the roots, later backfilling all that tasty real estate left behind. And how would you program the AI to look for strategic locations like choke points or “Panama Canals” and the like?

Anyway, I just couldn’t help thinking (and then writing) about this. Everyone seems to be cursing the AI as being stupid for setting up cities doomed to long-term failure, but I don’t see it as being so stupid. Overly optimistic, yes. But in several games I’ve taken cities from the AI that I wouldn’t have built in a million years that wound up being quite productive indeed. Had their empires expanded to link up with those cities instead of mine, the AI’s decisions might not have looked quite as foolish.

[kailhun]

I must agree that I'm not sure if this landgrabbing is smart or stupid. You often deprive a larger culture of a couple of squares of his land because any beginning city seems to have the 3x3 grid.
Many of these hopeless cities hang around a long time. If you weren't the human player the surrounding civ might fall and the motherciv could use the hopless cities as fallout bases. I had cities on the other side/middle of civs as well (part of a negotiated peace) and with the cultural strength of my motherciv these proved to be a nice way to nibble at the larger empires.
Although it was really annoying when the romans build a city right on top of the road I had build through the jungle connecting my original egyptian cities to the conquered greek/russian part.

Robert

[Th0mas]

Originally posted by Venger
Moraelin has some very good points - the AI is not only eminently beatable, but in many ways less of a challenge than the Civ2 AI. I don't know how much you guys have played the game, but the AI blows...an utter inability to defend itself, an inability to recognize it's own self interest, an inability to trade properly, ad infinitum.

I consider the Civ3 AI at almost an alpha level release level - there are too many substantive problems to consider it beta level...

Venger

Then your going to have fun with CTP2..

[benjy]

i think the AI is generally very good.

the issue i was discussing is not an AI problem. its a problem, no not a problem maybe thats too strong, its a result of the implementation of the border system; it allows gaps to appear between city borders which in the real world wouldnt happen (obviously in "real life" influence and borders are not calculated by reference to points and terrain squares!!)

im not a game desgner, and i dont if its possible to get around this, its just my 2 cents worth.....

would it be possible to have a set border around say 4 cities built a certain distance from each other i.e. the terrain in between those 4 cities automatically "belongs" to that civ, regardless of culture???