Fact: Civ 3 Combat is Rigged

[Alex 14]

In my previous game I was playing on a pangea map with one small island and one large island, I started on the small one, and the other civs started on the other one. In about 1Ad I gathered up a legion of 40 horseman and sent them to attack the Romans, followed by the Germans. The attack was successful and I only lost a few regular horseman which were reinforced soon after, from my first invasion I gained 3 elite units that survived to the end.

But then I attacked Russia, all my attacks were fairly accurate, except when I attacked one of their cities with my elite horseman, they had 3 regular spearman fortified. Each elite horseman took down 2 hit points before loosing 5. effectivly I lost 3 elites in the same battle situation killing 2/3 of the enemy, with full health, then loosing 5 hit points, what a loss.

Anyway, if the combat model is what people say it is, then this situation is rendered impossible. Perfectly randomized combat, until I get to Russia were I loose all my elites in the same battle circumstances at the same turn. The odds of the happening are fairly high for 1 elite unit, very high for 2, and impossible for 3 in the same turn.

There has to be some other aspect to combat that has been overlooked, I have never seen anything like this happen until my last game.

[ElBone]

dude, w/all due respect, I see nothing wrong with this. Horseman against fortified, city dwelling spearman? I'd expect to lose that, and roundly - no matter how many elite horseman to be thrown at them.

GL tho.

bone

[wolf7]

10 Jag Warrior can take down a veteran Mech Inf. fortified in a (small) city. That's unreal!

Most of the Jag's don't do a thing to the mech inf, but then one will get lucky and take two hp's. I mean, they're using clubs!

[N. Machiavelli]

Originally posted by ElBone
dude, w/all due respect, I see nothing wrong with this. Horseman against fortified, city dwelling spearman? I'd expect to lose that, and roundly - no matter how many elite horseman to be thrown at them.

GL tho.

bone

It's not that fact that he lost the horsemen that bothered him, it was that he lost all 3 of them in exactly the same way. All 3 units dealt 2 damage at first, then lost 5 in a row. Like he said, it's not impossible, but if the system was truly, consistantly random, than what occured is about the same odds as getting hit by lightning.

The random number generator isn't broken per se, but is pre-set before every turn and uses 'streaks' to make it's own job easier and speed up the game.

[simwiz2]

i noticed it too in a recent game: the generator is not random. i have seen the take 2/3 hp, then lose 4-5 (whether a vet or elite) occur several times in a row also. i believe the combat is random to a point, then the computer decides it "cant lose" and the unit becomes invincible, killing the stronger unit. this seems to happen alot at desperate times for them, like the last unit in a stack is attacked, redueced to 1hp, then i lose all hp. then i attack again and the 1hp unit defeats my 5hp attacker! it was my cav vs their unfortified grassland knights, so i should have had an advantage (which i did until it got down to that 1 hp) no matter what i did, i could not kill that unit. after losing 2 elite cav and 4 vet cav to the 1hp knight, i just ended my turn in frustration, then that knight attacked and died.

[Dis]

the russians might have had more units than you. This is my theory in my combat probablility thread.

It seems to me some weird number is thrown into the combat calculation- and I think it deals with the number of units of the human and the ai.

I wish it relied solely on attack/defense and terrain/city/walls bonus'. I'm certain they use some funky number in their "rolls"

[Capt Dizle]

Wow guys. I think at this point that Civ3 combat has achieved urban legend status.

[barefootbadass]

Originally posted by jimmytrick
Wow guys. I think at this point that Civ3 combat has achieved urban legend status.

Yeah, people get really superstitious when they lose 3 units in a row.

[Karhgath]

You say it's rigged, but, then you deny it saying you were able to conquer a lot of cities beforehand. How is that rigged? I mean, if it was really rigged, you would have a hard time conquering any city. Just a big stroke of luck and a stroke of bad luck. I don't see you whining about how you conquered X amount of cities in a row, but you do when you lost 3 units in a row. =)

Also, check out
http://apolyton.net/forums/showthrea...threadid=37155

It's about 'random numbers', it will explains a lot of 'weirdness' you see in Civ3.

[sophist]

I'm guessing the Firaxis programmers just used the OS's (pseudo) random number generator. These tend to be tied to the computer's time, which is measured in milliseconds, though finer resolution is possible. Then it gets processed through a bunch of functions that all serve to make the result wander greatly from the original time. So even if the initial seed times differ by just one, after applying a half dozen of these functions, you can end up with some completely different random numbers. Blaming the number generator for this is unjustified, as for just about all practical purposes, this is "random enough." I mean, these types of random number generators are used to create the one-time encryption system used by Secure Sockets Layer (AFAIK), aka SSL or https. If it's random enough for Amazon to rely on it for hundreds of millions worth of dollars in transactions every quarter, I suspect it's good enough for civ3. If there's a problem with the combat system, it lies elsewhere. I'm not even convinced by your example that there is one. Your statement implies (sorry) a lack of a proper grasp of probability.

At least in Civ's 1 & 2 combat seemed more logical. I think this had to do with the fact that modern units had many more hit points than less advanced units. Why did Firaxis think it was a good idea to make hit points the same for all types of units? There is no way that a spearmen can withstand or inflict the same amount of damage as a Tank! By standardizing hit points for all units they have made combat less realistic/fun IMO. If Hit points were more realistic, then I think the pseudo random number generator would be fine as is. Does anyone else see this as a major problem? It really bothers me that modern units are so easily destroyed by inferior units. I wish the AI would just upgrade their units. I don't mind a tank losing to infantry (very plausible, the soldier could have laid mines), but to Cavalry(did not work in WWII for the Polish)? Games like Civ should at least mimic the world from which they are modeled.

[Karhgath]

sophist : First, your name fits nicely =)

Second, statistically, PRNG are 'correct'. However, in small ranges, it could be totally 'unrandom', if you excuse the bad wording. Statistically, the following list of 'random' numbers is correct, if max = 10 :

1
2
3
4
5
6
7
8
9
10

Try anyway you want, but this is statistically correct and goes along probablity correctly. However, it is not how us, humans, perceive randomness. If you ask anyone, they will all say it is not random, althought it could, technically, be a result of a couple of dice you threw, which should be random. Also, encryption and all needs only one number, and most LCG are giving different enough results with good distinct seeds that it is ok. When generating a Key for PGP for example, you must move your mouse a good amount of time so it can generate a seed based on your movements. It is way better to generate a seed than using time, but it is still using the same basic formula, the seed is just more likely to be distinct.

However, when using a LOT of random numbers, one after another, you quickly see the discrepency. In the long run, yes, it will be statistically correct, but the discrepency will be even bigger than conventional randomness(for the lack of better wording) with dice for example. So, using like 20 numbers of a list in combat can yields weird results, as much as it can yield correct results. Like I said, there is nothing wrong, it's just that it is not how we perceive randomness. Statistically, it is correct.

[FalloutBoy]

The other day a guy walked into a store, bought a lottery ticket for a buck, and won millions of dollars. Want to talk about impossible?

[MarkG]

i "love" it when people use the word "fact" in their thread titles....

[Expatriate]

I think the thing that confuses many people is that "randomness" does not mean you have nice cyclic periods changing between low and high results.

Try drawing a box on a piece of paper and ask someone to place 100 random dots within the box. Afterwards you will notice that the dots will be pretty evenly spaced out and almost arranged in a pattern even though you asked for random dots. If you had a true random generator place those dots you would end up with some clusters of close dots and some big empty spaces as well as some areas with evenly spaced dots. The dots placed by the human are not truely random, but the person propably thinks they because thats how most people percieve randomness.

In other words, random means just that: Random! Sometimes you will get results that you think are not random because they look like a pattern. If that didnt happen there would be something wrong with your random numbers.

[docken]

Actually this "streaky" combat is now part of my strategy! For example, during combat I have 'tester' units whose sole job is to attack and see what the result is. If I see that they seem to win quite easily, I will usually mobilise my force and go on the offensive- as more often than not these seeded results favour the attacker. I have won many battles this way.

It has reduced warfare to sacrificing crappy units until I get a seed that allows me to completely crush the enemy.

[Wrong_shui]

FACT: Sharks Kill people
FACT: Vending machines kill more people than sharks
FACT: Coconuts kill more people than sharks and vending machines put together.

What do you get from these statistics?

1. If your swimming in the sea and a shark comes at you, dont worry.

2. If you want a can of coke you better be prepared to die.

3. People in countries with cocnut trees laugh at sharks and vending machines coz they're wuss's compared to coconuts.

[Akka]

Originally posted by number6
At least in Civ's 1 & 2 combat seemed more logical. I think this had to do with the fact that modern units had many more hit points than less advanced units.

*cough*
Mmh, are you SURE that Civ 3 fight is REALLY less logical than Civ 1's ?
*cough*

How many HP in Civ 1 ? One ? Oooops...

I do agree that we should be able to set HP on a unit-by-unit basis, though.

[kailhun]

The problem with civ III combat is the subjectivity of combat experience by the player. The memory of combat is probably skewed: remembering the bad experiences but forgetting when you got lucky.

So taking this into account, I do believe the combat system is predictable. On warlord level, indians (me) attacking aztecs. After having run the gauntlet of jaq. warriors and taking the core of the empire, I'm winning the war.
Cavalry armies (three units) walk up to a city and attack. Now, every time the same thing happens. The first one knocks of one or two points of the pikeman, then retreats and the second one comes up who always kills the pikeman. I started to rely on this happening and with succes.
Now as I understand the system, this shouldn't be happening, should it?

I also hate super-units (they won't die!), especially since I don't get them .

Robert

[sophist]

Originally posted by Karhgath
sophist : First, your name fits nicely =)

Hey, them's fighting words! Although I guess I'm asking for it.

Regarding the other bit about small ranges seeming other than random, I think you mean that for the guy who doesn't get probability. I mean, you can't eyeball 5 results and say they're not random. You can never say something wasn't random for certain; you can just say with varying degrees of certainty. He has a completely incorrect conception of what randomness is.

Also, your subjectivity point is well taken.

30 Helens agree: Civ3 combat isn't rigged. It's a fact!

[Alex 14]

Computers can generate random numbers! Sure if your using '/'or '\' or 'Mod' in a mathamatical equation the computer cannot render a completely random number, but what about 'rand'?? A simple piece of code that can generate a random number, if specified this number will have an exact chance of appearing as any other number in the code. For example if you have 1-10 integars, the computer will have a 1/10 chance of generating 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10, unlike humans precieve random actions a computer is fully accurate and will have the same chance of choosing any number.

Therefore I'll go back to my point that computers can generate random numbers, differently to ploting random points on a sheet of paper, and so forth the Civ 3 combat system cannot be perfectly randomized. There is another explanation for the circumstances of my previous game, the combat system is most likely randomized depending on variables.

[Easy Rhino]

I recently was playing a version of another game that happened to have a "streaky" random number generator out of some standard library.

A weapon had a dudrate of 50%. However, in usage, I found myself with 4 hits in a row, then 8 duds in a row, then two hits, and 4 duds...

Civ3's combat is kinda streaky. Big deal. If you wanted a certain unit to always win, then you might as not have random results at all.

ER

[GodSpawn]

Originally posted by Karhgath
sophist : First, your name fits nicely =)

Second, statistically, PRNG are 'correct'. However, in small ranges, it could be totally 'unrandom', if you excuse the bad wording. Statistically, the following list of 'random' numbers is correct, if max = 10 :

1
2
3
4
5
6
7
8
9
10

Try anyway you want, but this is statistically correct and goes along probablity correctly. .

WRONG! There are a huge number of statistcal tests for the "randomness" of a sequence of numbers, and this would fail them.

See, for example, "ITL Bulletin: A Statistical Test Suite for Random and Pseudorandom Number Generators for Cryptographic Applications" at www.itl.nist.gov/lab/bulletns/bltndec00.htm

You clearly know not of what you speak...

I've said it before, and I'll say it again - random numbers and large sample sizes can yield strange, but statistically predictable results. To the mathematically illiterate, this is the stuff of urban legends (and badly drawn conclusions...)

[Hurry]

I very much doubt that the PRNG in Civ3 is streaky. And if anyone is going to disagree with me, PUT SOME PROOF on the table! I´ve heard enough of "I know", "facts" and "it is like this", without anything than a single example to proove it. If the PRNG would never produce streaks, it would be more than strange.

And Alex 14: try to read the posts about the Pseudo-Random Number Generator. It might give you an idea what we are talking about here.

Try for example this link: http://apolyton.net/forums/showthrea...threadid=37155

EDIT: Added link.

[rid102]

I very much doubt that the PRNG in Civ3 is streaky. And if anyone is going to disagree with me, PUT SOME PROOF on the table!

Well I'm glad that your unsubstantiated assertion has proved us all wrong there while at the same time going someway towards contradicting itself!

I don't know about you but I'm quite happy with the Civ3 PRNG being as streaky as bacon and rigged like a ship with very many sails.

What can we offer you in exchange for the city of Veii?