This game is so arbitrary, makes me quit.

[Jokeslayer]

it turns a game of poker into a game of Yahtzee. Just roll the dice, nothing you can do to affect the outcome.

Wat?

If you're comparing Civ combat to poker, it's pretty valid, but saying there's no luck in poker (which you appear to be, but I may be misunderstanding) is just weird and wrong. Good civ players (relative to the difficulty level they play at) still win overall more than they lose even if they get smacked around by the RNG the same way good poker players win over the long run. The skill in the civ combat is in giving yourself the best odds (using terrain bonuses, bombarding before you attack, not attacking across rivers if you can avoid it, not attacking with spearmen etc.) Yes, the RNG will screw you over in a few battles, but you still win the war.

 

[rysingsun]

Did you successfully lose the city after reloading?

i had the "keep random seed" thingie on

 

[Pyrrhos]

There are better combat systems than the one Civ I-III uses such as that of the Panzer General series.

 

[MAS]

I figured the great Lighthouse was worth the investment on a map like this, isn't it?

Sure! I personally don't understand what it has to do with the topic anyway...

I also know the game is based on percentages, but it just seems to buck the odds so often. I have been playing wargames for years and I have a good idea of what is and isn't probable. I don't know....something is funky with the numbers generated in Civ 3

.

If I had a dime for every time someone claims something like this on this forum...

Yet, I don't have the same experience in my games.
There are payers who play at SID difficulty, they sometimes kill 1000 AI units in one turn. And I haven't seen these players post serious complains about odd combat results either.
Not surprising: with smaller numbers, you are more likely to get odd results, with bigger numbers, you are more likely to get very close to the expected results.

is civ 4 like this as well?

Civ4 also uses odds, but it has different mechanisms and concepts, such as "first strike" or situational modifiers such as +50% vs meelee, or +100% vs non-gunpowder units.

They did make it so that a full health modern unit can not lose vs an ancient age unit (such as spear beats tank) but it can still be damaged, and then lose vs the next AA unit.

And i know the post about getting angry when you units luck out was sarcastic, but it bothers me just as much. it turns a game of poker into a game of Yahtzee. Just roll the dice, nothing you can do to affect the outcome.

It is closer to a game of poker, you can't influence the cards you are dealt but you can still end up winning more money than losing money over a period of time.
Or rather, its a game of backgammon. You can't influence the dice, but a good player ends up winning far more often vs a bad player anyway.

But Civ3 is not a chess game, no, thats true.

 

[Sweetchuck]

You might be suffering from cognitive error.

Stick with it. Play some games on chieftan, build lots of military units and attack the AI with that "stack of doom".

If things go south quick, start a new game. Pick only 1 or 2 rivals.

 

[Katyusha44]

Sometimes I aslo hate the rng. I had 32 cossacks slauthered by bowmen, longbowmen and spearman. And any city I captured flipped back to the babs.

Sometimes the rng is awesome. I had one chariot get me a leader, individually kill 9 warriors, raze a city and capture another one. 2000 bc

 

[Optional]

You can make the outcome of battles a lot more consistant by going into the editor and giving every unit double hitpoints. I don't find that makes the game better though, it makes the outcome of battles too predictable.

I believe I've read that in Civ Revolution every unit that's at least three times as strong as it's opponant automatically wins the battle. That might not be a bad idea. A certain randomness is alright, but in Civ III some outcomes can be a little too freaky, like what was it; two or three swords not winning against a single warrior? That would make me scream as well.

On the other hand I would like to see that less advanced units still have a chance. To whom do those less advanced units usually belong? The civs that are lagging behind. If you're already in dire straits, you can use a little luck every now and then.

 

[TheOverseer714]

If the superior tech unit always won, it would either be too boring or too frustrating, and no one would want to play it. Sometimes combat results can be annoying, but most of the time higher tech does win, we just get a little when it doesn't go our way. Civ3 is a strategy game, not a simulation, so there has to be some leeway allowing for strategy to trump technology. Otherwise, from Demigod through Sid levels would be nearly impossible to win, rather than just very difficult(at least for average players). I might complain about things about this game, but I'm still playing after 7 years, not like most other games where you play it through once and you are done with it.

 

[Bartleby]

I'd be interested to see specific examples of what you're talking about.

 

[MAS]

You can just create a scenario in the editor where you give the AI 1000 of the same units yourself 1000 of the same units, all ready to attack when you start the scenario. And then you play this scenario. You attack with all of your units vs all of the AI units, and at the end of your turn, you count your units left.

Calculate the expected win/loss ratio. Pass the scenario around so others can also do the test, everyone post them in a thread on this forum. (don't forget to turn "preserve random seed" off)

 

[timerover51]

You can just create a scenario in the editor where you give the AI 1000 of the same units yourself 1000 of the same units, all ready to attack when you start the scenario. And then you play this scenario. You attack with all of your units vs all of the AI units, and at the end of your turn, you count your units left.

Calculate the expected win/loss ratio. Pass the scenario around so others can also do the test, everyone post them in a thread on this forum. (don't forget to turn "preserve random seed" off)

That is sort of what I am doing, but without creating the masses of starting units. Thanks for the idea, MAS. As for some results, last night, I had a spearman in a town, not city, although I have boosted defensive bonuses of towns to city level in my scenario, take a tank down to one hit point until all of a sudden the tank got him. Spearman started out with ONE hit point after bombardment. I had something similar happen with a galley verses a cruiser and a galley verses an upgraded Ironclad. In these cases, galley took off half of the hits points and then was killed. As this is not happening with dromon on galley fights, or early units on early units fights, it appears that there may be some programming in the AI that produces these results, which you might not be able to overcome.

 

[Pyrrhos]

There is a difference between a set-up and a game, the AI's evaluation on the neccessary level of challenge for the human player. Right now I am playing one of the scenarios "The Promised Land" (England, Monarch). In spite of keeping an eye on diplomacy and trading generously but not stupidly (only Persia was annoyed, Spain & India were cautious and Carthage polite), four AI civs landed troops on my island more or less on successive turns and declared. Now backtracking and taking their respective naval movement into account the decision to attack me was reached simultaneously by the AI.

Could this happen in RL? Of course. But in the context of the game, the AI decided to impede my progress a.k.a. challenge me. Since then and as I beat off the challenge easily, polite Portugal and polite Russia have declared too. I've previously made the contention that the AI civs should be looked at as one superciv aligned against you, the human player, and been ridiculed for "cognitive bias". This one is rather hard to explain away as cognitive bias & more likely, the accusers are the ones suffering from cognitive bias.

And Meisen! Your observation about hp losses is something I see in this game too!

 

[MAS]

Actually, this is one of the things where they improved the AI in comparison to Civ2. In Civ2 the AI was far to likely to form a "block against the human". People complained about this, because it broke the "suspense of disbelieve" and reduced the value of diplomacy.
One of the design goals for Civ3 AI was to make them act more as if they each had their own goals, and each had their own personality.
Its still only a simulation. And they still didn't do it perfectly. Civ4 has even more improvements in this area, however, even in Civ4 some players found the limits.

 

[timerover51]

timerover51

What you have been seeing is similar to what I've seen and is one of the reasons I started recording the battle results (other reason is I'm testing modding ideas and I want to make sure my changes are not skewing the game mechanics in unwelcome directions). Over the years I 've read here how people have tried testing the rng by lining up units face to face and then recording the results. All claim their results match the expected ones. What I'm concerned about is that maybe there is some kind of intermittant "adjustment" to the results that either gets triggered by certain circumstances (or has a % chance to be) or just kicks in randomly. And this factor may need actual game playing over many turns throughout a game if it is some events that are triggering it. Playing the game my impression is the further ahead of the AI the player gets, the more fierce the AI units can get. Just lining up a bunch of units and going at it for 1 turn doesn't simulate gameplay and could be unlikely to trigger an intermittant "adjustment".

The overall loss results I'm seeing don't seem too whacked out one way or another so far in the ancient age results I've looked at, but damage seems to be disporportionately laid on player units. You example of ship damage is something I notice a lot in the later ages too, when there is a large disparity in the ships involved. Example, it seems that more ofen than not my 15o-15d mow and 9o-9d frigates take 1-2 hp of damage attacking 1o-2d curraughs and 3o-1d galleys. The odds calculator says that the mow has odds of 24/15% and the frigate 30/22% vs the curraugh/galley to take 1 hp damage, which seems lower that the roughly 50% rate I've observed. Theoreticly, most should not be taking any damage. But so far I've not seen an AI curraugh/galley sink one of my frigates or mow.

It does seem to kick in on an intermittent basis when there is a big difference in tech levels. I just do not notice it when the techs are close. You have also boosted your ships a lot more than I have, so I would be interested in more info as to what your values are. I have been approaching this sort of cautiously as to really boosting attack and defense values, as then I have to deal with artillery and air attack values on ships, which promptly carries over to the ground units.

As for having several civilizations hit me at once, that has happened several times, which is why I tend to play water maps with continent or archipelago, and a limited number of opponents. That way it is harder for them to hit at once.

 

[timerover51]

Greetings, Meisen, I will need to copy your list and then send you my comments through an email. Just a couple of quick notes, however. You should give the Dromon more carrying capacity, I go with 3, as the Byzantines used them a lot for troop transport throughout the Med, and they were quite seaworthy for a galley type.

Your battleship and cruiser are too close together. WW2 cruisers had a broadside of around 2,000 pounds from 8 guns, although US war-built heavy cruisers had a broadside of 9 guns firing 335 pound heavy AP rounds for 3,015 pounds. An Iowa-class battleship had an AP broadside of 9 guns firing 2,700 pound shells for a 24,300 pound broadside. Iowa's 16"/50 was rated equal to Yamato's 18.1"/45 gun. For bombardment purposes, the British figured that the 1938 pound 15" HE shell was the equal of six 6" rounds for suppression purposes. This would be the equivalent of one battery of field to medium artillery. The US 16" High Capacity round weighed 1900 pounds and was the equivalent of the English 15" HE shell. If you figure the artillery in the game to represent an 18 gun battalion, the battleship should have about 3 times the firepower. It would also have a much greater ability to destroy hard defenses.

Lastly, you should not have the ability to upgrade conventional to nuke subs. The cost of nukes should be a lot higher than conventional boats, and you should have the option of buying both types. The edge the nuke boats should have is speed to deployment areas, plus normally a deeper diving depth and the ability to stay underwater indefinitely. I put that in in the form of greater hit points for the nuke boat, and more speed, equal to that of a destroyer.

 

[Pyrrhos]

The combined odds for this AI victory works out at 1,803,189,548,160 in 128,207,383,273,000,120,320 or 1/711,006 or 0.00014%

0.00014% is clear and unequivocal proof of a rigged system and I think we can throw out the apologist excuse of "cognitive bias" once and for all. Well done Meisen!

 

[Chamnix]

Save please .

 

[Jokeslayer]

0.00014% is clear and unequivocal proof of a rigged system

Not unless 0.00014=0 and nobody told me. I'm always the last to hear about that kind of stuff.

I don't think Meisen's post provides enough data to be regarded as significant. Interesting, sure, but not proof.

 

[Pyrrhos]

The apologists strike again...

Look here mate, if I had that sort of luck, I'd win $ 50,000 on the pools every week. Just give it up and stop making yourself look ridiculous.

 

[MAS]

Repeating myself:

You can just create a scenario in the editor where you give the AI 1000 of the same units yourself 1000 of the same units, all ready to attack when you start the scenario. And then you play this scenario. You attack with all of your units vs all of the AI units, and at the end of your turn, you count your units left.

Calculate the expected win/loss ratio. Pass the scenario around so others can also do the test, everyone post them in a thread on this forum. (don't forget to turn "preserve random seed" off)

I would have done it myself, but I'm on a Mac, and the Mac version of Civ3 doesn't come with an editor, so I can't.

I suggest a very simple and straightforward setup that uses very large numbers. 2 scenarios:

#1
Give the AI 1000 standard regular mace men in one stack, and the human 1000 standard regular spears in one stack on flat terrain, use standard rules, put both next to each other and let the AI attack.
Set it up it so that the AI can only chose to attack your stack of spears or not attack at all.

At the end of the turn, Just count the spears you have left.
count how many regulars have 1 HP left
count how many regulars have 2 hp left,
count how many regulars have 3 hp left,
count how many have become vets,
count how many vets have 1 HP left
count how many vets have 2 hp left,
count how many vets have 3 hp left,
count how many vets have 4 hp left,
count how many have become elite.
count how many elite have 1 HP left
count how many elite have 2 hp left,
count how many elite have 3 hp left,
count how many elite have 4 hp left,
count how many elite have 5 hp left,

#2
Same scenario, except that now the human has 1000 mace and the AI 1000 spears. The human does the attacking this time. At the end of the turn, Count the same things for the AI spears. See if the results are different.

You don't even have to calculate the odds with this, just see if the AI gets different results as the human.


Now post both scenarios as an attachment on this forum and let others run both tests as well, Then we will all combine our test results.

(Make sure its not a saved game with the "save random seed" option active)


The results format are posted like this:

Example:
        Human defends:  AI defends:
1hp reg 100              100
2hp reg 100              100
3hp reg 100              100
1hp vet  50               50
2hp vet  50               50
3hp vet  50               50
4hp vet  50               50
1hp leet 25               25
2hp leet 25               25
3hp leet 25               25
4hp leet 25               25
5hp leet 25               25
Total:  625              625

(obviously 375 of the 1000 spears died in these examples,
but thats not important to know.
I'm also not interested in what happens to the mace,
if you are, then give the mace men their own chart.)

This overview can show in one quick eyesight if there is a difference between AI results and human results.