doc mabuse
Warlord
@ Punkbass: not single units, two armies, so six units
My Elite tanks defeated by a veteran pikeman
- Thread starter doc mabuse
- Start date
- Status
Erik Mesoy
Core Tester / Intern
punkbass: Nice point 
Just skip someone up to Sid level and watch them complain that their spearmen aren't beating the enemy tanks. 
Ivan: Good idea, really.
Turner: *looks up at Post #95* what language? I said "clueless" once, and "terrorist" specifically as a caricature.
Just skip someone up to Sid level and watch them complain that their spearmen aren't beating the enemy tanks. Ivan: Good idea, really.
Turner: *looks up at Post #95* what language? I said "clueless" once, and "terrorist" specifically as a caricature.
This language:
Swearing is prohibited by the site rules. Getting around the autocensor is also frowned upon.
Erik Mesoy said:
Swearing is prohibited by the site rules. Getting around the autocensor is also frowned upon.
Erik Mesoy
Core Tester / Intern
Oh. I thought you would have moderated it earlier. And if that's frowned upon, I have a forum called Off Topic full of that.
tR1cKy
taking over the world
Again on the "3 elite legionaries beaten by regular spearmen" stuff.
For Tomoyo: first, thank you
. But the chances of this event are very small. Let's resume: 3 elite legionaries attacking a town on grassland with no walls defended by regular spearmen (i assume defendant units are fortified). According to Thunderfall's combat calculator, the spearman has a 0.186 probability to win (or 18.6%). Since i had 3 consecutive defeats, the probability is 0.186^3=0.0064 (0,64%). Again, this is too much random for me.
For Erik Mesoy: you can shove your sarcasm where the sun doesn't shine. You may have posted over 2,000 times in this forum, but those numbers don't allow you to mock people who simply post here their thoughts. You may need to learn something about "respecting other people" and stuff like that. Apart from that, your arguments are questionable. I'm referring to this sentence:
"Forcing random numbers to stay within, say, one standard deviation is the only thing out of line here. You can't."
There are your words, right? Well, you can. You say you're a developer. Ok, I AM a professional developer, and i can tell you that coding a mechanism which forbids a sequence of random numbers to go outside a predefined standard deviation is a simple task even for an average coder. All you need is a math library (which every language has) and a few minutes of spare time.
For Tomoyo: first, thank you
. But the chances of this event are very small. Let's resume: 3 elite legionaries attacking a town on grassland with no walls defended by regular spearmen (i assume defendant units are fortified). According to Thunderfall's combat calculator, the spearman has a 0.186 probability to win (or 18.6%). Since i had 3 consecutive defeats, the probability is 0.186^3=0.0064 (0,64%). Again, this is too much random for me.For Erik Mesoy: you can shove your sarcasm where the sun doesn't shine. You may have posted over 2,000 times in this forum, but those numbers don't allow you to mock people who simply post here their thoughts. You may need to learn something about "respecting other people" and stuff like that. Apart from that, your arguments are questionable. I'm referring to this sentence:
"Forcing random numbers to stay within, say, one standard deviation is the only thing out of line here. You can't."
There are your words, right? Well, you can. You say you're a developer. Ok, I AM a professional developer, and i can tell you that coding a mechanism which forbids a sequence of random numbers to go outside a predefined standard deviation is a simple task even for an average coder. All you need is a math library (which every language has) and a few minutes of spare time.
Cu Chulainn
The Unoriginal
- Joined
- Jun 21, 2004
- Messages
- 1,377
doc mabuse said:
Umm, I've never seen numbers like this you are either exaggerating or you have the worst luck ever. 70 bombers for 2 units? 15-20 for one? I've never seen this... EVER. I take cities defended by 3 rifles with 15arty and 6 infantry regularly, sometimes I get unlucky but overall my results are consistant. If I get the chance this week I'll run the numbers and see exactly how rediculously unlikely those situations are.
doc mabuse
Warlord
Believe me, if it happened to me once, i wouldn't have started this thread. The bombers: more than half come back home with a bombing failure. As said before, i attacked a city from the sea, 5 carriers with each 4 stealth bombers, 4 turns to kill a spearman and a horseman or cavalry, i'm not sure. What i do have to admit is that is was an own created map and modded game, in which the hp of a veteran was 15 hp. Still, more than half my bombers came home with failures, with a record of 7 failures in a row.
Want some more stuff? What about 3 ICBM's hitting one city, with 6 units: 3 units dead and no improvement destroyed (i still had a spy in the city, so i checked it).
The very first game i played, i was already in modern era before i started a war, I gathered about 20 bombers, all veteran, (this is a normal game in chieftain lvl), about the same in tanks, about 6 transports loaded with marines and started to attack a country.
Several things started to happen: before, my privateers could win from a galleon or galley, now a destroyer was attacked and destroyed. In an average amphibious attack of marines of cities without any coastal defence i lost a minimum of 2 marines per attack. I had one carrier in place with 4 bombers, no exaggerating here, in 3 turns, i made 3 hits and NO kill.
Next to a conquered city i had an elite tank fortified and a radar post next to it. A cavalry attacked me from the other side of a river, it destroyed me, losing one hp itself. No need to tell it took me hours to conquer not even half the nation. But i'll tell you anyway: it took me about 2 hours to conquer two metro's and six towns, while I am way ahead in science and upgraded units.
But I just started a new game, on chieftain lvl (just to check), only to see if it was an ordinary stroke of very bad luck or not. This time i will not stop before the whole world is destroyed! And i'll let you know what my findings are.
Want some more stuff? What about 3 ICBM's hitting one city, with 6 units: 3 units dead and no improvement destroyed (i still had a spy in the city, so i checked it).
The very first game i played, i was already in modern era before i started a war, I gathered about 20 bombers, all veteran, (this is a normal game in chieftain lvl), about the same in tanks, about 6 transports loaded with marines and started to attack a country.
Several things started to happen: before, my privateers could win from a galleon or galley, now a destroyer was attacked and destroyed. In an average amphibious attack of marines of cities without any coastal defence i lost a minimum of 2 marines per attack. I had one carrier in place with 4 bombers, no exaggerating here, in 3 turns, i made 3 hits and NO kill.
Next to a conquered city i had an elite tank fortified and a radar post next to it. A cavalry attacked me from the other side of a river, it destroyed me, losing one hp itself. No need to tell it took me hours to conquer not even half the nation. But i'll tell you anyway: it took me about 2 hours to conquer two metro's and six towns, while I am way ahead in science and upgraded units.
But I just started a new game, on chieftain lvl (just to check), only to see if it was an ordinary stroke of very bad luck or not. This time i will not stop before the whole world is destroyed! And i'll let you know what my findings are.
Erik Mesoy
Core Tester / Intern
And then the standard deviation shrinks
Not a predefined limit on numbers, but streaks of numbers at the edge of probability is what I meant. Unless you use www.NoEntropy.org to generate your numbers, there will be streaks at the edges, and one standard deviation (unless my norwegian math vocab does not translate) is supposed to include 67% of the results. And 5% of the results will be outside two standard deviations. (of course, if you use something like a coin flip with the values 1 and zero, huge rounding errors appear...)I'm not the developer in charge of the RNG. I program scenarios and write dialogue.
0,64% ...mmm... (100-0,64)^(100 civ players)=52,6% chance of it not happening to someone.
Across 1000 players, the chance is 0,16% of an event of this likelyhood never happening.
Do you know how I deal with bad luck? Overcome it. Shell all enemy units to 1hp and come in with armies when possible, overwhelming force otherwise. Bad luck has nothing on 6-to-1 odds. (two cavalry against a rifleman in a size <7 city.) Take one city at a time so I can concentrate my firepower. When an army loses 3/4 of its hitpoints on a bad streak, that isn't 3 tanks gone any longer. It means 3 turns spent healing before the same army marches out in case of other bad streaks.
doc mabuse: Can you post some savefiles, preferably with preserve random seed ON? It would be interesting to see these results as reproducable.
OBTW; Google returned 246 matches for what I used to "bypass the auto-censor". Feel like moderating them all?
doc mabuse
Warlord
@erik mesoy: i have no more old saves left but i will postone when it happens again in my new game
Cu Chulainn
The Unoriginal
- Joined
- Jun 21, 2004
- Messages
- 1,377
Found your problem, Crazy high health is going to make EVERYTHING take ridiculous amounts of units and time. Bombers only do a maximum of 3 damage per turn. AFAIK ICBM's have limits on this too. You seem to want to take all variablity out of the game, which would lead to mechanical play and zero chances of come from behind wins. I personnally would rather there be variation in my games. If you know how a game is going to go before it happens why waste your time playing it?
doc mabuse
Warlord
@ Cu Chulainn: I didn't actually, someone advised me to triple or max out the hp in the editor and i did. What i didn't know that it was applied to opponent as well, otherwise i would indeed not have bothered. And now i don't mod anymore. Cos despite everything i say here i (now anyway) like to play with a challenge. So, ok, concerning the modded game i take my righteousnous back but for the others, non modded, off course not
The reason why civ 3 combat is utter crap, is because they took a huge step backward from civ 2. In civ 3, for every combat round, ALL UNITS can only inflict 1 hp of damage. In civ 2, they had something called 'firepower' which determined how many HP of damage a unit could do per round. WHY they decided to take out this feature is beyond me. So in civ 3, a warrior that gets a 'hit' against a tank, will do the SAME amount of damage as a modern armor who gets a 'hit' against a tank. Granted, a tank will get more successful 'hits', but when you have 4 hp units, a warrior only has to get lucky ONCE out of 4 rounds, to take 25% health right off. Hell, if bugger gets lucky twice, that's 50% health taken off the tank. Yes, the tank will most likely win, but the current combat system is about 3 hitpoints away from simply being "all or nothing". So with the 'firepower' feature taken away (can't even mod it in with the editor), you'd think the developers would have some sort of reason for this. I don't know about you, but a battleship on my coastal waters no longer invokes an "oh crap" on my part (like in civ 2), but now simply makes me feel "ugh, now I have to waste a worker to rebuild roads, how tiresome".
Gee whiz. Fun for the whole family. I just *love* it when developers leave out features that were already present in past games and call it golden. Don't feed me poop and call it a sweet sweet brownie, because I'm not biting. However some people here eat up the feces like its some sort of prize, then try to convince others that the poo is in fact a brownie. Gee thanks... youz ah winnah!
Gee whiz. Fun for the whole family. I just *love* it when developers leave out features that were already present in past games and call it golden. Don't feed me poop and call it a sweet sweet brownie, because I'm not biting. However some people here eat up the feces like its some sort of prize, then try to convince others that the poo is in fact a brownie. Gee thanks... youz ah winnah!
michael4000
The Culture
All right, I'm back in.
Grav, after reading this whole uncivil thread, I know a lot about what details of the combat system you don't like, but not why you are so steamed about the net effect. You have said several times that for a technologically inferior unit to defeat a technologically superior unit is not "fun." Why not? Surely you accept that Civ is a game of chance, right? Because it sure would be a yawn if it wasn't. That being said, why are you upset what the odds are between any two given units? If you know the game, if you play it well, you know the odds going in, right? So: you attack the spearman, you take your chances.
You told me lo those many posts ago that you didn't care about historical accuracy, but I just can't see any other reason why it would bug you that one game piece (tank) could on rare occasions be beaten by another game piece (spearman). Any other reason.
So, I have to suggest again that it is very, very relevant that any reading of 20th Century military history will produce dozens of examples where far technologically inferior forces won battles, tactical victories, against far technologically superior armies. That's what the spearman -vs- tank thing is all about: it's a metaphor, like every other single aspect of the game.
Come on. Didn't you see Black Hawk Down? Just because you are ahead in tech doesn't mean you will be able to control every battlefield, everywhere in the world. That would be boring, and it would be wildly unrealistic.
Cheers all,
michael4000
Grav, after reading this whole uncivil thread, I know a lot about what details of the combat system you don't like, but not why you are so steamed about the net effect. You have said several times that for a technologically inferior unit to defeat a technologically superior unit is not "fun." Why not? Surely you accept that Civ is a game of chance, right? Because it sure would be a yawn if it wasn't. That being said, why are you upset what the odds are between any two given units? If you know the game, if you play it well, you know the odds going in, right? So: you attack the spearman, you take your chances.
You told me lo those many posts ago that you didn't care about historical accuracy, but I just can't see any other reason why it would bug you that one game piece (tank) could on rare occasions be beaten by another game piece (spearman). Any other reason.
So, I have to suggest again that it is very, very relevant that any reading of 20th Century military history will produce dozens of examples where far technologically inferior forces won battles, tactical victories, against far technologically superior armies. That's what the spearman -vs- tank thing is all about: it's a metaphor, like every other single aspect of the game.
Come on. Didn't you see Black Hawk Down? Just because you are ahead in tech doesn't mean you will be able to control every battlefield, everywhere in the world. That would be boring, and it would be wildly unrealistic.
Cheers all,
michael4000
The problem is, you cannot take out the whole random effect.
Civ3 is not like Chess.
Changes to the Combat System are a better thing to discuss for Civ4:
Take in mind that the wished thingie "modern units have REALLY good chances never to get defeated by a spearman" can be counter-productive on higher difficulty levels, where the AI often has a huge tech lead in the beginning.
One other thing is interesting, too:
3 defeats in a row, 0,64% chance? This is surely annoying.
I am not sure, but people have tested the Random Number Generator, and it works. So what is it that such things seem to happen that often?
Selective Bias - I remember only the odd results, much better than the normal ones?
Pure Bad Luck?
One should also take in account that random luck is most important in ancient eras: Few units, very similar attack values. A three-units-in-a-row-lost streak can be absolutely devastating.
There is a fundamental difference here:
Some would say "but it should not be impossible for that to happen!"
others "it should never happen!"
How about adding another button to preserve random seed? This one takes check of your negative results in combat, and if they exceed a certain level it rerolls to keep you within the mentioned range of 1 standard deviation, eliminating extreme values.
Adding HP (40 instead of 4 HP...) are only playable with animations off probably, but they can even out things a lot, too.
This would lead to the following results:
If you have the tech lead, you have more of an advantage than today. If you are a backward civ, a much harder time. I guess this would make Sid near impossible and Deity and Demigod - with all other settings unchanged - even more challenging.
But as they do not even bother the Sub bug and so on, this whole debate is pointless and should be lead in the Civ4 suggestions subforum. I still fear they already have implemented the combat model and that all suggestions will come too late probably.
Civ3 is not like Chess.
Changes to the Combat System are a better thing to discuss for Civ4:
Take in mind that the wished thingie "modern units have REALLY good chances never to get defeated by a spearman" can be counter-productive on higher difficulty levels, where the AI often has a huge tech lead in the beginning.
One other thing is interesting, too:
3 defeats in a row, 0,64% chance? This is surely annoying.
I am not sure, but people have tested the Random Number Generator, and it works. So what is it that such things seem to happen that often?
Selective Bias - I remember only the odd results, much better than the normal ones?
Pure Bad Luck?
One should also take in account that random luck is most important in ancient eras: Few units, very similar attack values. A three-units-in-a-row-lost streak can be absolutely devastating.
There is a fundamental difference here:
Some would say "but it should not be impossible for that to happen!"
others "it should never happen!"
How about adding another button to preserve random seed? This one takes check of your negative results in combat, and if they exceed a certain level it rerolls to keep you within the mentioned range of 1 standard deviation, eliminating extreme values.
Adding HP (40 instead of 4 HP...) are only playable with animations off probably, but they can even out things a lot, too.
This would lead to the following results:
If you have the tech lead, you have more of an advantage than today. If you are a backward civ, a much harder time. I guess this would make Sid near impossible and Deity and Demigod - with all other settings unchanged - even more challenging.
But as they do not even bother the Sub bug and so on, this whole debate is pointless and should be lead in the Civ4 suggestions subforum. I still fear they already have implemented the combat model and that all suggestions will come too late probably.
Sarevok
Civ3 Scenario Creator
For some odd reason, ive never had that happen to me. I train my tank crews well...
Erik Mesoy
Core Tester / Intern
To paraphrase;
You dislike the combat system,
therefore it is poop.
Others like it,
therefore they are self-deluding lying poop eaters.
Based on the popularity of Civ, I would rather say that we ARE being given brownies, and you're a paranoid contrarian who tries to convince the rest of us that the brownies are inherently bad and should be replaced with what I will term "oatmeal porridge". Bland, flavorless, and predictable.
*me whacks a 10-kilo stone axe into a tank, wonders how many warriors make up 1 unit, and decides that a warrior can justifiedly do damage to a tank*
@Longasc: Like I posted above, the chance of a 0,64% event NOT happening is the minute 0,16% among 1000 Civ players. Expect the unexpected.
tR1cKy
taking over the world
Erik, your statement is correct, but pratically there are many workarounds. Solutions may be very simple or fairly complex, can involve minutes or hours (not days!) of work and can solve the problem partially or completely, but all of them are easily implementable by an expert coder (as those at Firaxis' team should be).
Let's consider these events:
A) my tank lost to a spearman.
B) my elite legionary lost to a regular spearman fortified in a town with no walls.
C) the event B is replicated 3 consecutive times.
Here's some possibile ways of solving the problem:
1) assuming the RNG produces floats in the range (0,1), let be T a threshold level (say 0.01 or 1%). If the generated number is less than T or more than 1-T the number is discarded. With this simple method you cut out extremely lucky or unlucky rolls, but this doesn't assure you that events like (A) won't happen at all, they are only made less probable. You simply cut the edges of the gaussian bell.
2) Let E be the random event and P(E) its probability function. Let T be the threshold level (as above). If P(E) is less than T, the event won't happen (no rolls are made). If P(E) is more than 1-T the event will surely happen (again, no rolls are made). This method is able to prevent events like (A) and (B) from occurring, but won't stop an event like (C), which is a sequence of events extremely improbable in its whole.
3) Let E, P(E) and T as above. Let S be a work variable which initially has a value of 1 (100%). This variable will be used to store probability of past events (or sequences of them). This is a little complex trick, but easily understandable with an example.
Let's take T=0.01, or 1%, as above.
a) my first legionary lose to a spearman. Chances are 18.6%. S takes the value:
S*0.186=1*0.186=0.186.
Now S>T. The event is allowed to occur and my legionary is killed.
b) my second legionary lose again. Chances are again 18.6%. S now is:
0.186*0.186=0.035.
Once again, S>T and my second legionary follows the first in heaven
c) my third legionary lose. Chances are the same. S now is:
0.035*0.186=0.0064.
Now S<T. Event is forbidden, the legionary wins despite the roll and S is reset to its initial value of 1.
That's all.
Let's consider these events:
A) my tank lost to a spearman.
B) my elite legionary lost to a regular spearman fortified in a town with no walls.
C) the event B is replicated 3 consecutive times.
Here's some possibile ways of solving the problem:
1) assuming the RNG produces floats in the range (0,1), let be T a threshold level (say 0.01 or 1%). If the generated number is less than T or more than 1-T the number is discarded. With this simple method you cut out extremely lucky or unlucky rolls, but this doesn't assure you that events like (A) won't happen at all, they are only made less probable. You simply cut the edges of the gaussian bell.
2) Let E be the random event and P(E) its probability function. Let T be the threshold level (as above). If P(E) is less than T, the event won't happen (no rolls are made). If P(E) is more than 1-T the event will surely happen (again, no rolls are made). This method is able to prevent events like (A) and (B) from occurring, but won't stop an event like (C), which is a sequence of events extremely improbable in its whole.
3) Let E, P(E) and T as above. Let S be a work variable which initially has a value of 1 (100%). This variable will be used to store probability of past events (or sequences of them). This is a little complex trick, but easily understandable with an example.
Let's take T=0.01, or 1%, as above.
a) my first legionary lose to a spearman. Chances are 18.6%. S takes the value:
S*0.186=1*0.186=0.186.
Now S>T. The event is allowed to occur and my legionary is killed.
b) my second legionary lose again. Chances are again 18.6%. S now is:
0.186*0.186=0.035.
Once again, S>T and my second legionary follows the first in heaven
c) my third legionary lose. Chances are the same. S now is:
0.035*0.186=0.0064.
Now S<T. Event is forbidden, the legionary wins despite the roll and S is reset to its initial value of 1.
That's all.
doc mabuse
Warlord
This is an argument that has been pointed out numerous times: the whole game evolves around development and science. So can you tell me the point in putting effort to do so the fastest if the inventions you make (which in this case upgrades of military) have no great effect on other nations. Then why bother to put in military upgrades at all if you can't count on an additional power?
I believe in the case of Black Hawk down we're talking of not even one whole regiment of marines against a whole town, with modern weapons as well and not naked spearman. How long did it take the USA to take in Afghanistan? And to make a measure: how much longer will it take to take in Irak (once again they're not up against spearman).
If you want to come up with Afghanistan vs USSR, don't, cos then we're talking about minor military attention of an army completely demotivated and old tanks against mounted gunmen, which is again not comparitive with ancient naked howling spearman defeating a whole regiment of tanks.
If it were just a case or rolling the dice then why bother puttting time, money and effort in a "more efficient army"?
Ask Bush, i'm sure even him will get the joke in that.
I believe in the case of Black Hawk down we're talking of not even one whole regiment of marines against a whole town, with modern weapons as well and not naked spearman. How long did it take the USA to take in Afghanistan? And to make a measure: how much longer will it take to take in Irak (once again they're not up against spearman).
If you want to come up with Afghanistan vs USSR, don't, cos then we're talking about minor military attention of an army completely demotivated and old tanks against mounted gunmen, which is again not comparitive with ancient naked howling spearman defeating a whole regiment of tanks.
If it were just a case or rolling the dice then why bother puttting time, money and effort in a "more efficient army"?
Ask Bush, i'm sure even him will get the joke in that.
punkbass2000
Des An artiste
Your argument is flawed. They do have a great effect. Losing 1 tank out of 100 to a spearman (if that) is not 'no great effect'. If you don't believe that technolgical and economic superiority will almost certainly yield a win, I would be happy to play you in a PBEM game.
- Status
Similar threads
