Difficulty level should not be done by CHEATING!

Mavtse

Chieftain
Joined
Jun 23, 2002
Messages
14
I think Firaxis had done a big wrong thing in the way it set difficulty. Difficulty level should be done via the power of AI, not by cheating.

for example, in western/chiness chess game, stronger AI difficulty is determined by the ability to estimate more and more future turns.
In Civ3, it is impossible, but in other way: AI ability in handling various aspect of the game(worst algorithm to the best).
In this way, it is still a fair game. Yet, the Civ3 now is not fair game because it cheats.

Cheating i suspected(not in editor):
1) AI knows player's city status and troop information without the need of spy
2)Attack/defence bonous, making ridiculous battle result

So what do u think?
 
NiceOne made a great thread about cheats, supected, proved and false... Try search it, all is there !
And just for the record, programming a CIV3 Ai is a lot more difficult that a chess AI, because the rules are far more complex and various... And victory is not only one condition... And it's already huge to plan in advance every possible moves in chess, just think of how impossible it would be for Civ, with not only moves but also building, outcome of battle, and so on.
So even playing on algorithms would be extremely time -consuming, because to define one is long, and I think there are quite a lot of Algorithms in CIV3

So I think you're a bit unfair to Firaxis... :) My reproach to them would be that they do not make the AI cheats known.
 
Originally posted by Mavtse
for example, in western/chiness chess game, stronger AI difficulty is determined by the ability to estimate more and more future turns.

It's not possible to make AI for Civ3 to play as Chess. In Chess, there are 32 pieces to for the computer to think, yet at the most difficult level, it takes the computer many days to make a move. If they do that in Civ3, we could wait for eternity for the AI to move.
 
If you take away the Ai cheating aspect you'll have a game that is far too easy.

TAke Shogun:Total War for example, they removed the "horde" problem induced by the AI being allowed to go -ve money. The game went from being extremely hard to extrememly easy overnight.

At least if the game is hard it provides a challenge. Too easy a game just gets shelved.

I'm happy with the way civ3 difficulty levels are worked. :) Paly at regent if you want a fair game. To program different Ai levels for each difficulty level would be a worthless exercise. I'd rather they spent the time making the best Ai they can for all levels, then use other means to aid or hinder the human (as they have done).
 
All AI programs cheat. It isn't realy cheating, its programming. In Civ it gets bonuses the "human" doesn't.
In first person shooters there are more monsters or the monsters are tougher. Is that a cheat?
It isn't possible with current technology to make a truly smart AI.
 
Well... On Deity The AI starts with 2 or 3 settler's 4 Warriors An archer (depending) And A spearmen or two. And yes it is possible to make Smart AI, it has been done, In games, although I have never seen smart AI In a strategy Game.
 
There is a reason you haven't seen a smart AI in a strategy game.

People already complain about how long the AI turns take in the late game in Civ3 as it is. Imagine the howls of protest if the AI was smart, and had to take an hour or so per AI Civ per turn. Even with the fastest machines currently on the market, a smart AI would simply take too long running through the varous permutations of smart strategy.
 
On deity, the AI starts with 2 settlers instead of 1 and 12 military units (8 offensive, 4 defensive, I think).

Yes, you could program a smarter AI for each difficulty level....But then expect to pay $500 for the game!
 
Thanks Bamspeedy, 12? hmmm maybe I was thinking of a different level? oh well, And yes Padma I agree, I would much rather have a quick game than a Long Drawn out Game where the AI kicks my ass every time. This way I can do much better by realizing that Im fighting the Computer, not a human, so it follows the same path, And most likely, wont start a war over Iron, Like I do. ;)
 
I got it backwards. It's 8 defensive units and 4 offensive units. An extra settler, and 2 extra workers. All of this is in the editor.

Back to the original poster, yes the AI does know where your units are, but combat results are the same throughout all levels. Except for battles against barbarians, as they get harder as the level gets higher (I think this applies to humans only, not sure). Also, on higher levels the AI gets more unit support/city.
 
According to the editor, the bonuses against barbs apply to AI civs too. A possible way to slow down initial AI expansion is to remove these bonuses so that the AI loses more units and settlers to barbarians attacks.
 
I don't consider the AI advantages on higher level "cheats." But, on the topic of AI advantages (and specifically barbarian combat bonuses):

There's another less-well known AI advantage at the higher levels - the editor setting labeled "Default AI Difficulty Level." This is normally set at "Regent."

The AI difficulty level does not make a "smarter" AI -- it imposes certain penalties / bonuses on the AI in various areas. Think of it as choosing, for the AI, its own difficulty level -- just as you choose your own difficulty level at the beginning of the game. The three principal factors that are confirmed (by Soren) to be affected by this editor setting are: (1) AI combat bonus against barbarians; (2) number of citizens born content; and (3) likelihood of good / bad results from popping goody huts. (There may be additional advantages / disadvantages - I've never really thought about it extensively, but would assume that any other factors / variables that affect the human player by moving from difficulty to difficulty would also affect the AI by playing with the editor setting).

Set the AI Default Difficulty Level to "Diety" in your mod and your AI opponents will have no bonus against barbarians, will only have 1 citizen born content, and will get less favorable outcomes from poppoing goody huts. Set the level to "Chieftan," and you will have a real challenge on your hands - and AI immune to barbarians, popping great huts, and with absolutely no happiness problems! (Note that by defaulting to Regent, the difficulty level works just as the production bonuses / penalties apply to the AIs - the human player has an advantage when playing below Regent, is the same at Regent, and is at a disadvantage above Regent).
 
Originally posted by Mavtse
I think Firaxis had done a big wrong thing in the way it set difficulty. Difficulty level should be done via the power of AI, not by cheating.

for example, in western/chiness chess game, stronger AI difficulty is determined by the ability to estimate more and more future turns.
In Civ3, it is impossible, but in other way: AI ability in handling various aspect of the game(worst algorithm to the best).
In this way, it is still a fair game. Yet, the Civ3 now is not fair game because it cheats.

Cheating i suspected(not in editor):
1) AI knows player's city status and troop information without the need of spy
2)Attack/defence bonous, making ridiculous battle result

So what do u think?

A lot of us are sick of AI cheating, including AI civs ganging up on the human.

In Civ 2 combat odds changed as we went up in difficulty level. At least it was up front, and not sneaky like this AI.

I said before and I say it again, playing against this AI is like playing against an evil ******ed person.
 
Originally posted by PCHighway
And yes it is possible to make Smart AI, it has been done, In games, although I have never seen smart AI In a strategy Game.

Your right! I have inside information that Firaxis actually has a really good AI that is smart and is able to learn. Those bastards at Firaxis have been hiding from us. It's a highly parallel program and running on the secret server farm buried deep within the Firaxis bunker. It's in the middle of the 4th turn now, but with advances in computing technology they hope to complete one game before the heat death of the universe.

If you didn't notice the sarcasm, what I'm trying to say is that you CANNOT have a "smart" AI that learns from it's mistakes the runs on commerical hardware. An AI for anything much more complicated than checkers that can actually keep up with humans is not going to run fast enough for anyone to actually play. Perhaps I'm wrong, but I can't seem to find an example of "good" AI. Can anyone?
 
Originally posted by provolt


you CANNOT have a "smart" AI that learns from it's mistakes the runs on commerical hardware. An AI for anything much more complicated than checkers that can actually keep up with humans is not going to run fast enough for anyone to actually play.

Everybody keeps repeating this mantra: The hardware is not fast enough. And this opinion is based on what? You don't know of any strong AI programs? You are going to exclude chess because it is simpler? The way I look at it is that, both in chess and civ3, there is a "horizon" beyond which a human being can calculate. I don't see why it's not possible to make a stronger AI for civ3, one that can calculate somewhere in the neighborhood of the human's "horizon". It's going to be really difficult to prove that something like that could not be done. You should propose a program's design before you decide what the hardware requirements are.

But I do see why it's not worth doing for Infogames/Firaxis/whoever. The time and money spent would be too costly. It might take 10 years or more to do, and in the meantime everybody would hate them. The players would say, "where is our next release?" The investors would not take 10 years to fire everyone in the company. If there were truly a big demand (from people willing to pay) for a strong AI they might be able to hire more people to do it faster. Might. That strategy doesn't always work.

I don't think it's the hardware. It's the cost. IBM was making operating systems that ran in 256K of memory 30 years ago. It was expensive to do, but they could make a profit.

I think the reason that you don't see more strong AI programs is that software companies couln't make money if they developed one.
 
First of all, good one, Provolt !


Originally posted by sumthinelse
It's going to be really difficult to prove that something like that could not be done.


I dimly remember you can not prove something can't be done.It does not make sense.
But I can't prove it :)
 
Originally posted by Mullet Crusader
Yeah, its impossible to PROVE something CAN'T be done :)

If you could prove that you needed 1,000 terabytes of memory to make a stronger AI, then I would say, OK, I accept your proof. Most people who buy civ3 don't have that much memory. But the fact that most of the commercial AI programs do not play well does not begin to convince me that the AI can't be strengthened.

I would agree with you that it would be hopeless for Firaxis to try to make an AI as strong as the chess programs that run on home computers, but for a different reason: Not because it's impossible, but because it's too expensive.
 
Top Bottom