civ 3 is not a fair and good game

[Catt]

Originally posted by jobuck
hmm... okay
This is very interesting.
I have to confess that I didn't had really take care of these A.D.M parameters.
I have to try playing the game taking that stuff in consideration.

Everyone has their own tastes and Civ 3 won't be for everyone. But it would seem a shame to throw out a game that many, many people find enjoyable simply because of some early frustration -- otherwise you may miss something you will come to really enjoy once you understand it, and you presumably wasted your money (assuming you bought the game).

Those "A.D.M. parameters" are absolutely fundamental to understanding how combat works in Civ 3. If you're playing the game with no concept of how A.D.M affects what happens within the game, it's probably a pretty safe bet that you don't yet understand any of the dozens (hundreds?) of other important gameplay concepts. From your posts, it sounds to me like you could quite easily decide that the card game Bridge is stupid, unfair, and not a good game . . . because you wouldn't bother to pay attention to those funny pictures, colors, and letters / numbers on the cards and their impact on the game, and can't understand why you keep losing.

I'm not trying to hit you over the head, but Civ 3 is not a simple, straightforward game. If you don't make a minimal effort to try and understand some of the very basics of how the game works, you're doomed to frustration (and deserve it too!).

Catt

[Louis XXIV]

Basically, Civ3 is a strategy game.

Every unit has a role.

I'll try suggesting a few tips (keep in mind, I'm no expert)

When you start with a Civ, you get a settler, a worker, and (if you're expansionist) a Scout.

Each square around you has 3 things:
Food (When your cities "Food box" fills up, your population will grow)

Shields (When you build a unit, it requires shields. If a unit requires 10 cost, like a warrior, it will require 10 shields to be built)

Commerce (Commerce increases the amount of money you have. It is important to invest this money in science. In the Domestic advisor, you can invest up to 100% of your money into science. This might be advisable, because it will give you techs quicker. If your citizens become unhappy, you can put that money into luxuries)

You probably know about exploration, since you played. Units with 2 moves move faster. however, mountains, hills, forests, and jungles slow down this movement. If you go on a hill or mountain, you get to see further.

Now a little more complicated strategy.

Early in the game, the more food you have, the better. You want that food box to fill so you get more population. When you get a second person, you get to put him on another square, and get more food, shields, and commerce. When you get to about 3 or 4 people. Build a settler. This will remove 2 of those people, but give you another city. This city will allow you to build more units, and get tech faster.

Workers improve the terrain and make it more productive. Roads make units move faster and add extra commerce to the tile you're working. Irrigation allows extra food, mining gives you shields.

I could go into much more detail, but that's what the manual, civlopedia, and just playing the game is for. If you ask, I (or somebody better at the game) will give plenty of tips

Here's a discription of the units. All units have their stats written as cost (in shields), attack, defense, and movement

Settler 30 0.0.1 You use these to build cities. If enemy units attack them, they capture them. You can do the same.

Worker 10 0.0.1 Improve the terrain. They can be captured

Scout 10 0.0.2 fast unit available to expansionist civs. Can be captured

Warrior 10 1.1.1 cheap scout early in the game. Not good for defeating anything but other warriors

spearman 20 1.2.1 the ideal defender. It has the best defense of any unit, and has a fairly cheap cost.

Archer 20 2.1.1 cheap cost, but poor defense make this unit "OK"
Its real advantage comes with the fact that it, like spearmen and warriors, requires no resources (see swordsman)

Swordsman 30 3.2.1 expensive for the early part of the game (downright cheap compared to later units) swordsman have the best attack and defense of any unit. However, they don't get produced as quick. They also require iron to build. This means that you can't build them without a road connecting the iron resource to your city.

Chariot 20 1.1.2 As you might see. this unit isn't good at attacking spearmen (what most cities will have). It is quicker than other units, but it requires "Horses" to build. Horses work like iron. If attacked, it has a chance to retreat before being destroyed

Horsemen 30 2.1.2 although this unit has less attack then a swordsmen, many people prefer it, because of its speed. It also requires Horses

Catapult 20 0.0.1 This unit is a special unit with a "bombard" feature. Instead of attacking units, it bombards them (use the "B' key). It can't take damage, but it will only do 1 hit point of damage (It has a 4 bombard, meaning it will have a 4/(4+D) chance of winning. If it is a spearmen defending, it has a 4/6 or 2/3 chance of doing 1 damage out of 3, 4, or 5. Bombard will never destroy a unit. This has its advantages, but catapults aren't the best bombard unit) It can be captured, so keep a unit with it

That is all the info you need about units, until you enter the Middle Ages.

Hope that helps

[statusperfect]

Originally posted by Olaf Hårfagre
Eat and die...

that is not very nice...

cheers

[Tiberius]

jobuck, you should definitely spend a few hours reading, a few hours that will save you a lot more hours of frustration. Civ3 is a great game and if you have spent already lots of hours playing it, you would probably love it, once you catch its little "secrets".

Here's the link:
"Must Read" threads for newer players / posters

There is nothing to read there , but a bunch of links to other threads, in different categories, written by expert civ players, advices of what, how and when to do, etc.
The info is priceless and you can't find it in manuals. For example:
Winning Early: What do YOU do?
or Civ choice for early warmongering
or Civ choice for the "builder"
or Combat System Explained
and a lot lot more.

One more thing. You said that a good strategy game is a game you can predict. Then think a little bit: a good strategy means that you plan a campaign, prepare for it and then accomplish it (like capturing that strategic city over there, near the iron resource). Strategy is not about winning or losing a battle, but about winning the war. Sh*t sometimes happen, just like in real life. This game is a recreation of human history, and just like in real life, a knight not always defeats a pikeman. There are tons of other factors that can (and did) change the outcome of a battle: terrain, supply lines, courage, morale, determination (like somebody defending his own homeland, correct?), leaders, tactics, etc. However the civ that has the best military, the best developed infrastructure, most friends and allies, control over resources, best strategy and tactics, best leaders and highly traines soldiers, etc; well, generally wins (the war).

[nbarclay]

jobuck, are you talking about a genetic algorithm type of approach? Even with infinite computing resources, almost all the testing would have to be done AI vs. AI because games against human players take so long. And AIs that do fantastic against other AIs might have fatal weaknesses that human players could identify and exploit to win relatively easily.

Further, Firaxis's computing resources are far from infinite. At one second per turn, averaging 300 turns per game, games would average five minutes each. Even if we assume 1000 games per computer per day (a little over three times that rate), and 100 computers running for 100 days, that's only ten million test runs. By genetic algorithm standards, with as many "genes" as it would take to represent a Civ 3 strategy, that's a joke. The only way it even might be possible to get adequate computing power would be with a SETI@home type arrangement, and I'm skeptical that even that would be enough for more than minor refinements with how complex Civ is.

Then there's the problem of figuring out how to define your "genes." A given civ's strategy at any given point in the game needs to be based on the status of the other civs in the game, especially its neighbors. How do you define "genes" with enough complexity to represent that, and how do you make an evolutionary process work on such "genes"? I suspect that answering those questions would take more time than it took to develop the Civ 3 AI - and would be far more likely to result in failure.

And as if that weren't enough, Firaxis is not a scientific establishment that's purpose is to provide the most effective AI players possible. Rather, it is a commercial game company with the mission of producing a game that is as much fun as possible for as wide a range of players as possible. Thus, AIs have to be optimized not just for their effectiveness in winning but also for how much fun they are to play against for various types of players.

It's interesting to contemplate how a company with infinite resources might tackle the problem of developing a first-class Civ AI. But the reality is that Firaxis had to deal with designing an AI using limited manpower and limited computing power in a limited amount of time. They're a commercial company, and past a certian point, investing more resources in the AI just to win a tiny handful of additional sales is a losing proposition for them.

Maybe someday competition, advances in computing power, and advances in AI technology will converge to produce AIs that can compete with the best human players on even terms in games as complex as Civ. But for the time being, we have to settle for rules that give AIs built-in advantages on higher difficulty levels to compensate for their inferior strategy.

Nathan

[Panag]

hi ,

if it aint a fair or good game the move on and buy something else , .......

have a nice day

[Dominae]

Originally posted by nbarclay
jobuck, are you talking about a genetic algorithm type of approach?

Actually, I'm the one that threw the genetic algorithm idea into the discussion, with my example of the Checkers player (the name eludes me right now). But the point remains, namely that learning algorithms are pretty tough to implement for something like Civ3.

You could avoid genetic algorithms by defining an extremely complex evaluation function for each turn (or each component of each turn). Doing this, and getting it just right, would be a mammoth task, definitely far more complex than jobuck's simple example (wx1+wx2+...).

Further, Firaxis's computing resources are far from infinite. At one second per turn, averaging 300 turns per game, games would average five minutes each.

Actually, they could probably reduce the time-per-turn considerably (much less than one second). But again, the point remains: it would take much too long for the program to learn anything.

They're a commercial company, and past a certian point, investing more resources in the AI just to win a tiny handful of additional sales is a losing proposition for them.

I agree. To be the World Champion in Chess, you'll probably have to resort some esotoric algorithms and computing. Just to get to the same level of competence as the Civ3 has now would require a huge initial investment if neural nets were used. Why go through all that, when 99% of Civ3 players get a run for their money with an AI constructed by (more or less) one person?

Maybe someday competition, advances in computing power, and advances in AI technology will converge to produce AIs that can compete with the best human players on even terms in games as complex as Civ. But for the time being, we have to settle for rules that give AIs built-in advantages on higher difficulty levels to compensate for their inferior strategy.

Actually, I believe that without resorting to learning methods, it should at present be possible to design a Civ3 AI that trounces human players. It's just that a small gaming company like Firaxis will not come up with such an AI (nor would they, or any other company, want to afford to). It's a cost/benefit issue, pure and simple.


Dominae

[jobuck]

Hi nbarclay, Dominae

I don't know how much familar your are with neural network, but if you are the following models are useful to assist a decision:
retropropagation, boltzman or kohonen

Such model are particulary useful in such situations:
- the rules that could allow to solve a problem are unknow or very hard to formalize.
- we have some example of input and ouput that the model should provide (the output being given by some subject matter expert)
- the problem can change with the time (e.g. the initial condition)
- the problem request high speed of treatment

So, this seems to me to fit with the problem of creating an ai.
And as I mention, it is just here to assit, not to provide THE solution.

jobuck's simple example (wx1+wx2+...).

There is a missunderstanding here about what I say, as it is compared against genetic algorithms.

But the dummy "formula" above have nothing to to with ai. So their is no needs to compare it against it.
But if we take a look to some expression of the kohonen model, we have something like Mi(t) = Mj(t-1) + a(t) * [E(t) -Mj(t-1)] - formula updating the weights
This is still a linear equation, easy to compute (even if a high volume of iteration is needed)
That's why I say above that such model can be used when "the problem request high speed of treatment".

So how to integrate neural network into ai ?

The *spontaneous* idea was to first define a model, providing the input. The hard-coded ai then provide the neural network with such value and ponderate his decision with the result he get.

What is important here, is that the neural network (NN) don't was to simulate the whole strategy of the game.
We can have for e.g. one NN who simulate the trading, one other that simulate the fighting and so.
We can imagine also another model who take as input the result of the previous model, and so on until get a orientation for the next step to play.

Another point is that we can imagine model that anticipate 1, 2,3,5 or the 10 next "move".
But at each turn, the prediction restart with the "current state" of the game.

Anyway, that was a spontaneous idea.
It comes to my mind because I used such model years ago. And as mentioned, NN may solve complex problem (such as produce weather forecast).

I don't know if you remember in the 80's when the japanese were arguing that before the year 2000, they will have produce the first intelligent computer, able to communicate with human.
Despite the high level of technology that master the japanese, that still not the case today.
And it's a very complex problem that no one have some today.
But one way that produce the "less bad" result in may domain is the NN.
I was supposing he could have a role to play in such game.
This need to be investigated to provide a better "feasibility" answer.
But that's just a suggestion.
As this domain insterest me, I would perhaps try to work on that.
But don't worry. If I get at the end only 50% of the capacity of civ3, I ll already be happy.
Now my comments regarding civ are at another level, more general.

[jobuck]

panag:
Please do not come with political pix here, or choose another thread than my post.
Because sharon is definitely not a clean guy for me.

[Panag]

Originally posted by jobuck
panag:
Please do not come with political pix here, or choose another thread than my post.
Because sharon is definitely not a clean guy for me.

thats apolyton jobuck , .....


are you going to complain about everyones avatar that you dont like at poly , .... good luck

[jobuck]

panag,

it's up to you to keep it as your avatar, as it's up to me to comment it.
If that's apolyton then take my comment as apolyton as well.

Note:
I have nothing against the people this guy represent.
But as a bloody settler, that guy don't have my respect.
I don't care on who care about my opinion; again, I express it freely, that's it.

[Sirotnikov]

jobuck,

Wish to come to the Off Topic and debate about the supposed bloodiness and settlement activity of Sharon?

[/shameless plug]

ON TOPIC:

Things just cost too damn much. I rarely rushed cmopared to Civ II. And diplomacy is waaay too costly. Not as costly as espionage though.

It's cheaper to either buy the tech, or build and army, and conquer a city, than it is to steal a tech

[jobuck]

King,

I don't want debate this point.
The only opinion I agree to express about this is
I respect your avatar and your flag.
Final point.

[Cruddy]

Just to add 1 point Jobuck - about trading with other Civs.

If you want military alliance (or anything else the Civ has like a new technology) better to put it on the table and ask other Civ what they want for it.

Sometimes they will say no deal at all - very rare they will want nothing for the swap either.

Usually you have to pay other Civs for alliance (it costs money to fight wars). You can also trade techs, and luxuries, and strategics. Making other Civs your friends is not easy - but making them your enemy is easy.

Actions you do can make other Civs annoyed or Furious - more likely to declare war on you.

I would not say Civ3 is not fair or a bad game - but it takes many lost games to understand your bad play.

I've played it for over a year, and there are still many things I am not good at.

Don't let Panag upset you. He will bump any thread with a silly comment because he is like that.

I must admit, I have also done this sometimes!

[TOUGHUSAGUY]

LIFE AINT FAIR! GET USED TO IT PEOPLE!

ABOUT SHARON?? I SUPPORT HIM ALL THE WAY! GO GET THE SAND **** AND GIVE ALL THE THEM WHAT THEY DESERVE! WAR OT TERROR MEANS IT LIKE THIS !

USA USA USA USA USA USA USA USA USA USA USA USA!!!

Added later by Ming... And another DL bites the dust