Getting sick of the AI's inaccurate unit strengths

Urshulgi

Chieftain
Joined
Feb 22, 2002
Messages
3
Few situations i've come across in my crusade against the japanese.

1) A veteran samurai with 2 health successfully defends against, and destroys a 3 elite cossack army that started with 12 units of health. The samurai was in a hill fortress and the army was attacking from a plain.
2) A veteran samurai with 2 units of health defends against and destroys two full health veteran marines in a row. They were both in hill terrain.
3) A veteran calvary unit successfully defends against a veteran TANK. calvary unit was in a city and tank was attacking from plain.

So I thought that maybe the computer was just using bull**** to defend itself against me. Then the japanese launched their offensive.

1) Calvary unit routs a dug in infantry unit from a city. The calvary unit was attacking from a plain.
2) Veteran samurai destroys a fortified veteran rifleman in a fort on a mountain. The samurai was attacking from a plain.
3) Samurai defeats a veteran infantry unit, both on a plain.


For a game that they tried to make historically accurate, you would think that units with MACHINE GUNS would annhilihate dudes with swords. And you'd think that TANKS would kick the hell out of horses. It's pretty annoying to have to use overwhelming technologically advanced forces to defeat the computers archaic units.

Hey Sid Meier, here's a clue. Machine guns > swords. Tanks > horses.
 
Here we go again.:cry:
 
Originally posted by Urshulgi
Few situations i've come across in my crusade against the japanese.

1) A veteran samurai with 2 health successfully defends against, and destroys a 3 elite cossack army that started with 12 units of health. The samurai was in a hill fortress and the army was attacking from a plain.
2) A veteran samurai with 2 units of health defends against and destroys two full health veteran marines in a row. They were both in hill terrain.
3) A veteran calvary unit successfully defends against a veteran TANK. calvary unit was in a city and tank was attacking from plain.

So I thought that maybe the computer was just using bull**** to defend itself against me. Then the japanese launched their offensive.

1) Calvary unit routs a dug in infantry unit from a city. The calvary unit was attacking from a plain.
2) Veteran samurai destroys a fortified veteran rifleman in a fort on a mountain. The samurai was attacking from a plain.
3) Samurai defeats a veteran infantry unit, both on a plain.


For a game that they tried to make historically accurate, you would think that units with MACHINE GUNS would annhilihate dudes with swords. And you'd think that TANKS would kick the hell out of horses. It's pretty annoying to have to use overwhelming technologically advanced forces to defeat the computers archaic units.

Hey Sid Meier, here's a clue. Machine guns > swords. Tanks > horses.


Get over it. Civ III is a game, and the units are just pictures with numbers attached to them. A samurai attacking an infantry has a 4/(4+10) = 28% chance of winning. Not so unlikely. Don't think of it in terms of "this guy has a gun he should win". Think of it in terms of numbers (cuz that's how the computer works it out... the computer doesn't care whether the pretty little picture has a gun on it or not).

- Windwalker
 
Aside from Windwalker's response, you also have to consider other things... while unlikely, a large group of samauri warriors COULD kill riflemen. You have to consider the possibility of a surprise attack or the samauri using some kind of unconventional method to fight which gives them the upper hand.

As for cavalry defeating tanks, tougher to justify, but it's unlikely that'll happen except very rarely. Also, don't look at the attacker's terrain, it means nothing. All that matters is the defender's terrain. So your first three examples aren't quite as unlikely as you think. I mean, in the first one, the samauri was on a hill (+50% def) AND in a fortress (another +50% or maybe more? Can't remember...). That makes a big difference. And the veteran/elite status also means nothing. It adds to max HP, but does nothing else.
 
Aside from Windwalker's response, you also have to consider other things... while unlikely, a large group of samauri warriors COULD kill riflemen. You have to consider the possibility of a surprise attack or the samauri using some kind of unconventional method to fight which gives them the upper hand.

a battle in civ3 happens over the course of a year, so those samurai would have to be pretty lucky, as in waging a long term successful campaign against the riflemen, because in civ terms the samurai aren't getting lucky the combat system means they have a decent probablity of winning...i think it's better to fix the combat system than to make excuses for it

basically civ3 just need to double hitpoints and to raise the industrial and moder era units stats slightly

And the veteran/elite status also means nothing

except for radically altering the chances of victory ;)
 
Of all my many complaints about Civ III, combat resolution is very low on the list. It is really not so bad in my experience. Could be improved, perhaps.



BTW, it is spelled CAVALRY. "Calvary" is where Christ was crucified.
 
The few historical occasions where horses came against machine guns or tanks ended in massacres for the poor equestrians.
 
Unless the game designers put an automatic defeat rule in people will run into instances where unlikely results occur. For instance, if I read the original poster correctly, s/he had already increased the hit points of the cossack unit at least. It still lost to that fortified samurai. I'd say that a player whose got a big lead in tech is going to have hundreds of mismatched combats in a standard game. Mismatched assaults by the player of tanks against muskets or weaker, mismatched counterattacks by the AI of knights or cavalry against those tanks. Sooner or later the improbable happens, and sooner or later a string of improbable events occurs.

I'm surprised I haven't seen this suggested before, but there could be a surrender rule. If the chance of defeat is greater than 90%, the weak unit surrenders. I don't know how to resolve that... does the winner get the unit? POWs cost money to feed and all. If the average Civ 3 player had to make a choice between micromanaging the expense of feeding a pain in the ass artificial construct or killing it, we'd commit the atrocity. So it would probably be better if the unit just disappeared.

Maybe the reason why it hasn't been suggested is because it's a bad idea, but jeez, it would cut down on these constant posts about odd combat results.
 
Blah blah blah... :rolleyes:

The computer doesn't give AI's ridiculous combat bonuses against the player, even if ot does in other parts of the game (research, production etc). For every 'surprise' loss I had at one point of the game I had an equal amount of 'surprise' victories at another point of the game.

Learn to live with it or shut up. :lol:
 
Panda's right...the odd results go both ways...I once had a caravel that survived 2 attacks by veteran ironclads. I guess the stringy nature of the combat generator is what I don't get. Being a Mac user, I'm new to many of these old arguments, but I find the random combat generator to be far from that sometimes. Out of curiousity I once tried combat with 5 different ironclads against enemy ironclads. I would fight one battle, reload the game and fight another battle at another location...the results were exactly identical...no matter which ship I attacked. I read somewhere in the endless threads on this subject that Firaxis did this to help stop the save, fight & reload if you don't like the results trick. I think however that they have hurt the true random nature of the combat. If someone wants to exploit the fight & reload trick instead of playing ironman style, let them...it was their $40.
 
What your are seeing Rustus Maximus is that the game is simply saving the seed for the random number generator with the save game. This should not have any effect on the random number generator at all. It will just ensure that the the strings generated from that point forward are the same.
 
Thanks for the condescending and arrogant comments. Go back to eating paint chips.
 
Well, I sure am glad I responded. Go back to getting your dumb ass kicked by the AI then, champ.
 
Originally posted by Rustus Maximus
the odd results go both ways...
Not always. I'm getting pissed at the random number generation as well. I'm not going to whine with specifics, but in my last war I lost despite having numerical superiority, technological superiority, and I was fighting from fortified defenses most of the time. It didn't go both ways, I was waiting for the entire 20 turns of the war for the 'its going to go right eventually' to happen, and it didn't, and my numerically superior army was defeated again and again, and I lost miserably, and I'm mad about it. What IS the point of the little numbers if they are SO random? :mad:

Now I'm executing all my generals, disbanding all the underperforming units that aren't already dead, and moving to an island in the middle of the meditaranian until I can build a single ICBM and launch it into the Chinese capital a few hundread years later for revenge.
 
come on folks, don`t be that abusive! It`s unfair, especially when directed against newbies or people with little time who have played few games and gotten a real bad string.....

I`ve seen good strings and bad strings and equal strings of luck, and I have to say: they are too long! When I win a war fought with inferior numbers, inferior tech and bad territory, then go on the offensive with archers against musketmen, later Knights against Riflemen and WIN, total being like 30 funny combat results (say: bigger then 7:1 odds) in maybe 80 fights - I`ll be pissed! Even though I win! It`s just stupid when trying to good and succeeding gives you the same chance to win as trying to be good - and failing miserably!

Also, it is sooooo frustrating to finally have gotten into the game, to know what to do, when to do it, plan for the unexpected, doublecheck, calcutalte the odds again and again, finally commit to the attack - and loose again and again and again!

I don`t play CivIII so I can wait for a lucky break to win!!!!!


So please, be kind to those who are really down from these experiences! The next patch should include a HP doubling or trippling IMHO; then rare things happen rarely!
 
is the 1.16 ver better than 1.17 ? and why? as a new player to the game one of the things that bothers me the most is the way archers can kill off my riflemen and even if it were the reverse my archers killing their riflemen, its still not right and very unlikely. why bother pushing my R&D people so hard to discover gunpowder,tanks,aircraft if a group of archers can shoot down my Messerschmit fighters? (is this possible? havent got aircraft yet, that would hella-suck!) another thing that bugs me is that the years pass by way to fast it seems. its 2030 and i dont have a modern army, still using riflemen and cavalry units mostly. or am i just that damn slow. are you sure the next patch will include a better ballance for the combat units like HP doubling? that would be a great thing IMHO. hope it comes out soon.
 
I've played thousands of battles. The randomizer appear "true." Remember, in a collection of hundreds of players, some will be lucky, some will be sad. That is "random" also.
 
Well, based on my own experience and the experience several other people around here have reported, it does seem that the PRNG is not terribly good. Not that it favors one side over the other, but that the correlation between successive PRN's is too large. While overall it seems to all average out, the streakiness appears excessive. This is not, however, unexpected. The majority of PRNGs in use today (especially the default ones most compliers have) have this fault (or the opposite one, in some cases), because avoiding it takes many more calculations per number. It's a tradeoff between speed and better (i.e. closer to independent uniform distributions) randomness.
 
Originally posted by Beamup
Well, based on my own experience and the experience several other people around here have reported, it does seem that the PRNG is not terribly good. Not that it favors one side over the other, but that the correlation between successive PRN's is too large. While overall it seems to all average out, the streakiness appears excessive. This is not, however, unexpected. The majority of PRNGs in use today (especially the default ones most compliers have) have this fault (or the opposite one, in some cases), because avoiding it takes many more calculations per number. It's a tradeoff between speed and better (i.e. closer to independent uniform distributions) randomness.

good to read that, I was starting to feel like I`m the only one with that impression.....
 
Top Bottom