The Real problem with Difficulty Levels.

[Ision]

Firstly, let me say that CIV is my favorite game. In spite of whatever faults I may list here, in my opinion there is no other PC strategy game that comes close.

The fundemental problem with CIV is the difficulty levels. As you move up in difficulty the game becomes progressively narrower in your strategy options. The AI bonus does not create a better game on higher level. It merely makes the human become more of a number cruncher and ultra micro-manager.

All expert players of any games are great micro-managers. The differance heres lies in the fact that on the 2 highest levels of CIV - micromangement is THE game. After Monarch - it's a game of 'catch up and kill'. Period - end of story. The game loses nearly all it's nuances.

The reason for this is the manner in which the game was designed to become harder. Massive initial bonus's to the AI does not maintain the same game at a higher level. It changes the entire dynamics.

Everyone knows the drill on Regent or Monarch level. A somehat close first age. A good lead in the second age. A forgone conclusion the moment you land TOE. I only use this example as one of many. For some of you the game is over much earlier (like the moment you take out that neighbor CIV's capitol). As the game goes on the AI is progressively handicapped by it's inability to make adjustments to it's gameplay. The human's micromangement ability begins to take it's toll on the AI's hard coded inflexibility.

The solution in my opinion is clear. The AI bonus's should be 'progressive' in nature. Imagine for a moment an Emperor level that went like this: Intial production bonus 9 - 1 extra worker - 2 extra defenders - 4 free units - AI trade 140. As each AI civ enters a new age it's production bonus's continue to go up as well. Middle ages production 8, Industrial age production 7, Modern era each city supports 4 free units. Ect...

That's only an example. I don't want to argue the details. Naturally a better formula would be arrived at that maintains a proper play balance. There would be other tweaks as well. But the essetial element is a slower but steady AI bonus that coincides with what is typically the humans progressive lead with each passing age.

I maintain that an age by age 'progressive' AI bonus system would create greater play balance for the better players, without sacrificeing those aspects of CIV that we all loved when we were still new to the game.

The idea is to restore the flavor the game had when you won your first couple of games on Regent or Monarch, but before you mastered those levels and went on to a completely different 'CIV game' on Emperor and Diety

[Ision]

By the way, naturally each CIV will enter the new age at different times. This will only eccentuate the game even more. It creates even more killer AI's. As in real history, on average those civilizations that did not keep up technologically were doomed. Even if they were isolated from the rest of the world geographically, they were doomed on first contact with the technologically more advanced (as was the case with the Aztecs upon encountering the Spanish).

In game terms it will reduce the long stalemate type AI vs AI wars.

[dexters]

I think if Firaxis could somehow committ the resources and have a lone programmer work part time on the Civ 3 Ai over the long term, as a kind of comittment to the community, we could very well end up with a very challenging game with things as is.

The problem is, AI fixes only seem to come with the big bad patches. I was extremely disappointed when I heard that there would be no major AI fixes in PTW. Soren ended up giving us fans a treat by improving several aspects of the AI and continued to do so in 1.14 and to 1.21.

Conquests, with its theme of war etc., needs to address many of the AI issues regarding naval invasions, stack movements, stack management and address some of the cheap war related and non war related exploits that some of the players out there use consistently or perhaps even subconciously.

I would even go as far as saying that I would gladly pay a subscriber type fee to Firaxis for their continued support of Civ III beyond the mandatory patching.

Those who followed the game into PTW have esentially had their vanilla Civ3 patched since day 1 (oct 2001 to feb 2003), and I can attest to the great improvements in the AI even though fixing the AI has never been the focus of these patches. I guess part of me is hoping for some serious effort to beef up the AI with Conquests.

A commitment to patch the AI would greatly energize the community and the game as we mull over exploits and how best to help the AI put up a better challenge. Perhaps I'm one of the few people here who believes in the Civ3 AI and its ability to do great things if enough resources is devoted to adapting it to common human playing styles.

[I Am Jeff]

The backbone of Civ3 is the A.I., the better the A.I. is, the better the game will be. Ideally the only difference between dificulty levels should only be difference in A.I. stratagy. Simply meaning that higher difficulty levels, should only mean harder A.I. I doubt that will happen with Civ3, but maybe if they could continue and improve the A.I. greatly, maybe the bonuses could be lessend.

I too believe in the A.I. and the great thing Soren, I now laugh when I look at the A.I. of previous civ games. I know that if there are not great A.I. improvements included in the new XP, I may have trouble spending the initial $30 for it.

[Jawa Jocky]

good idea

[player1]

I remember that CtP2 has "progressive" AI bonuses.
The later in game you are the more bonuses AI has.

Of course, it didn't really mattered much since in vanilla CtP2 AI is the worst AI compared to any Civ-like game.


P.S.
Something similar happend in Civ2 too (the famous 1750).

[BulMaster]

I find giving the bonuses to the AI stupid. Really, the game becomes somewhat unfair. Before somewhat starts argueing with me, they can make the ai intellingent. I'm sure is not that hard, we see this happen in galactic civilizations. The AI doesn't get any real bonuses it becomes smarter as the esiest example when you mass your troops on the border you receive a msg what are you doing, just like in real world. Giving millitary bonuses aka free units is the also stupid. I understand to give bonuses to the AI if it was impossible to make it "smart", but hey we are the witnesses that it is possible i probably requires some more programming but as long as the public is willing to buy crap stuff why improve it. As I said in galciv no bonuses and the ai still manages to kick u at the hight lvls. But who cares, nobody listens to me anyway.


Kind Regards

[johncmcleod]

I completely agree. But as long as people will keep on buying Firaxis's games even without a better AI they won't spend all of that extra time and money to make it better. Bonuses are even worse in sports games. I'll be Italy in World Cup 2002 but on higher difficulty levels easy teams such as Ecuador will be much faster and better shooters than my guys.

[dexters]

To be fair, the Civ3 AI engine was written and designed(probably) in 1999-2000. If they had to rework the entire AI for today, it will no doubtedly be superior.

That said, I am not asking for an overhaul. Most people here aren't. We just want upgrades to the AI (to quote Neo from The Matrix Reloaded). The suggestion I'm making is an effort on Firaxis' part to keep patching the game (AI patches) as long as they can with Conquests.

And I concur with what was said earlier. Civ3 Conquests without major AI improvements would be really disappointing. I'd like to think I'm paying $30 for more than just new graphics and new units.

[sabrewolf]

Originally posted by BulMaster
Before somewhat starts argueing with me, they can make the ai intellingent. I'm sure is not that hard,

BulMaster,
I totally disagree! making the AI is the hardest part of the game. why?
because the computer doesn't have the parallel processing our brain has. so you've got to simulate strategy with an intelligent valuation system. e.g. my profit from researching military tradition is greater than music theory.
there are a lot of things to be considered and the more factors you pack into your algorithm, the slower the game gets (a lot of people dislike this fact too).

i used to play deity games, but lost or gave up 8-9 out of 10. simply because the AI starts with 2 settlers (which the human normally gets after quite a few turns). and the production bonus (AI needs 40% less shields). it's just too much for standard sized maps. i've given in up now. if the AI were better, these bonuses wouldn't be needed... but as we've seen it's not that easy.

however, some parts should be done for c3c:
- learn to use the luxury slider efficiently
- try beelining for techs (knowing the trade value of this)
- learning early rushing
- rating trade. a ROP shouldn't be that important if you don't have any touching borders...
and most important of all
- diplomatical intelligence. humans pick on the AI one by one. how about a worldwide alliance once you declare war on the 2nd civ after killing or badly weakening another one?

basicly these things are all exploits the human can use but the AI can't...

[djafrot]

I agree that designing the AI is by far the most difficult part of creating the game.

But I really hate the idea of just jacking up the AI's bonuses to make the game harder. It's hard to judge things when you can't equate what you've got with what your opponents have... i.e. "there's no WAY he could possibly have that many units!".

[ChickenHawk]

I agree that writing good AI is very difficult. So we are left with giving bonuses to the AI to keep it competitive. Perhaps by the time CIV VII comes out they won't have to do this any more.

I've played a few games on Emperor Level and was thinking I should try Deity after I finish my current game. But I'm kind of scared to try after looking in the editor at all of the extra bonuses they get; especially the extra offensive and defensive units. I have decided to create a custom difficulty level somewhere between Emperor and Deity.

I will start by only changing the "Cost Factor" from 8 (Emperor) to 6 (Deity) so that the AIs can build stuff much quicker. Ision contends that progressive improvements in the AI's bonuses is necessary to prevent us from running away with the game after holding out after the initial onslaught. I hope to find a happy balance where the opening attacks may not be quite as fierce (fewer bonus AI combat units, no extra settler, etc), but the middle game will remain challenging (low AI Cost Factor).

If I can win at this new level, I might try lowering the AI's Cost Factor even more, or maybe giving the AI an extra settler and/or worker (as in Deity).

Has anyone tried experimenting with these settings?

[sabrewolf]

chickenhawk: the hardest thing on deity is the extra settler and worker for the AI players.
that gives them a jump start you only have, if you get a settler from a hut (very rarely).
that's why winning on a smaller map on deity level is a lot more difficult than on a large/huge map. there you have time to catch up.

the units don't help much. havn't seen the AI exploring much at all in the beginning... and offensive units for non-militaristic civs are anyway only warriors. only militaristic civs get archers.

i quite successfully tried standard sized games with the standard deity-start without the extra settler and worker. made my 10% success rate into something around 40% ;-)

[dexters]

Originally posted by sabrewolf

the units don't help much. havn't seen the AI exploring much at all in the beginning... and offensive units for non-militaristic civs are anyway only warriors. only militaristic civs get archers.

You must not be looking. I see the AI exploring with their warriors all the time. they're just not as methodical in exploring as a human player and are prone to missing goody huts. It seems like their exploration is primarily to move in a direction in hopes to finding a new Civ.

But at least the AI doesn't cheat and go strait for goody huts.

[sabrewolf]

we know if we're going to expect a tundric region or a wide jungle. we try to step onto mountains to see more. i think the AI just tries to find yet undiscovered spots. so if he walks past a hut... bad luck...

yes, you're right, he does send them out. but i'd be sending out more - at least one of the defensive units... you can always send those the "shorter way" (e.g. check the coast you started at or head to the nearer pole to look for furs and game)

[ktaek]

Wow i think you guys are being harsh. The AI in Civ3 is why i keep coming back to this game, and it's a signficant achievement in game development.

Everything said is valid, the only change in difficulty should be better AI, but cars should run on water & air and we should be able to teleport to alpha centauri by now as well.

Consider how much more complicated Civ3 is vs Chess, and then think of the massive resources dedicated to creating Deep Blue.

And consider that so far there has never been any AI that has been able to be considered even passable at playing Go.

The amount of depth the Civ3 AI has and the fact that i bought it for about how 1/2 of what my polo t-shirt cost makes Civ3 the bargain of a lifetime imho.

[djafrot]

You're right... this is a phenomenally complex game. The AI often surprises me in its ingenuity (sp?) and makes it a different challenge every game. Admittedly I don't play as much as many of the others here so I am not as good...