I do not care what you say. (Part 2)
- Thread starter Bigv32
- Start date
wioneo
King
- Joined
- Jun 11, 2006
- Messages
- 752
The reason for this topic was originally to show how the rng was incorrect. This did not happen, and the topic creator seems to have lost interest in the topic or spontaneously combusted. In either case, there is really not much discussion value left here and it would probably be in everyone's best interest to just move on, because some bad juju is floating around and it's tearing our forum apart...
I can't find a "cryin face" smiley to exemplify my emotions so I'll end off with
I can't find a "cryin face" smiley to exemplify my emotions so I'll end off with
mboettcher
General
- Joined
- Aug 24, 2007
- Messages
- 524
The 'odds' displayed are the odds of winning. However the odds do not reflect the statistical distribution of how damaged the victor may be due to first strikes.
A unit with more first strikes vs a unit with more power but they both have the same odds in a combat situation will be very different results. The first strike unit is more likely to survive with less damage. Alternatively the same unit promoted down either the combat or drill lines will have very different results. The COmbat line will have better odds of winning but the drill line will have better odds of doing more damage in a loss (particularily vs lopsided situations) and a better chance of surviving with more HP.
Also Drillline tests require MANY more 'test battles' to accurately test the statistical distribution as there are 1-4 different numbers of first strikes available depending on the promotion level of the unit. Each number of first strikes is equal i probability so 4 different battle outcome distributions are 'averaged' for a D4 unit. This means that you probably have to test a D4 unit roughly n^2 times versus a combat line unit to get a similar idea of the distribution (in other words you have to smooth out both the 1/4 chances for different number of first strikes and then the wode ranging results of those first strikes).
The reason First strike units survive with more HP when they survive is that they are esentially weaker units fighting perhaps more powerful units with more HP. They have a lower chance to win their first roll or two but that may be all that is necessary to win a battle. Every time the lose a roll they take more dmg than a similar combat unit (different/worse power ratios).
It is probable that because power ratio's can really amplify the power of first strikes (as even parity means a great deal more dmg per strike and more dmg per a won round after that and more likely to win those rounds) while combat units get a diminishing return of effectivness after parity that defenders are better off taking Drill promotions do to multipliers offered (and the siege resistance).
Granted CG is a powerful defense line but against large stacks of artilery one is definately going to benefit more from Drill. Drill multiplies the damage reduction of having more units than an artillery can do collateral to in one round. If one has nine DIII longbows defending a hill city, and perhaps one or two are D4, then even 10 trebs will only do 25-30% total damage to each longbow
and the D4 longbows will take minor damage while probably killing all the attacking trebs. The result is that the enemy may have to spend many more hammers to take the city than you do defending as you will probably kill many of the attackers.
The MAJOR drawback of the drill line is the Knight immunity to first strikes. The era on high levels where drill becomes most relevant is the Ren/Middle ages and knights/Curs here are devestating as drill defenders/siege resistance vs pikes/mounted defense do not necessarily stack well and the pikes will take the most dmg from the artillery and rendering the drill longbows promos useless with already weak power ratings even weaker vs knights.
A unit with more first strikes vs a unit with more power but they both have the same odds in a combat situation will be very different results. The first strike unit is more likely to survive with less damage. Alternatively the same unit promoted down either the combat or drill lines will have very different results. The COmbat line will have better odds of winning but the drill line will have better odds of doing more damage in a loss (particularily vs lopsided situations) and a better chance of surviving with more HP.
Also Drillline tests require MANY more 'test battles' to accurately test the statistical distribution as there are 1-4 different numbers of first strikes available depending on the promotion level of the unit. Each number of first strikes is equal i probability so 4 different battle outcome distributions are 'averaged' for a D4 unit. This means that you probably have to test a D4 unit roughly n^2 times versus a combat line unit to get a similar idea of the distribution (in other words you have to smooth out both the 1/4 chances for different number of first strikes and then the wode ranging results of those first strikes).
The reason First strike units survive with more HP when they survive is that they are esentially weaker units fighting perhaps more powerful units with more HP. They have a lower chance to win their first roll or two but that may be all that is necessary to win a battle. Every time the lose a roll they take more dmg than a similar combat unit (different/worse power ratios).
It is probable that because power ratio's can really amplify the power of first strikes (as even parity means a great deal more dmg per strike and more dmg per a won round after that and more likely to win those rounds) while combat units get a diminishing return of effectivness after parity that defenders are better off taking Drill promotions do to multipliers offered (and the siege resistance).
Granted CG is a powerful defense line but against large stacks of artilery one is definately going to benefit more from Drill. Drill multiplies the damage reduction of having more units than an artillery can do collateral to in one round. If one has nine DIII longbows defending a hill city, and perhaps one or two are D4, then even 10 trebs will only do 25-30% total damage to each longbow
and the D4 longbows will take minor damage while probably killing all the attacking trebs. The result is that the enemy may have to spend many more hammers to take the city than you do defending as you will probably kill many of the attackers.
The MAJOR drawback of the drill line is the Knight immunity to first strikes. The era on high levels where drill becomes most relevant is the Ren/Middle ages and knights/Curs here are devestating as drill defenders/siege resistance vs pikes/mounted defense do not necessarily stack well and the pikes will take the most dmg from the artillery and rendering the drill longbows promos useless with already weak power ratings even weaker vs knights.
frekk
Scourge of St. Lawrence
So you're complaining that the combat odds are wrong, but you didn't bother to check them in your tests?
Ai Shizuka
King
- Joined
- Oct 30, 2008
- Messages
- 856
Oh, if you want more math humour, I'm sure I could find you tons more. With more spherical humour,
A mathematician goes to the racetrack with his friends, and they all casually bet on the horses. After the first few races, the mathematician's friends notice that he's dead on, getting each race correctly. They decide to ask him, "Hey, so what's your big secret to betting correctly?" The mathematician replies, "Oh, yeah, I devised this amazing formula for predicting the winner. I've been testing it out recently, and noticed it seemed to do very well." Intrigued about his methodology, his friends poked him a bit more. "So, how do you do it?" The mathematician then starts out explaining, "Okay, so first thing, I assumed all horses were identical and spherical..."
A mathematician goes to the racetrack with his friends, and they all casually bet on the horses. After the first few races, the mathematician's friends notice that he's dead on, getting each race correctly. They decide to ask him, "Hey, so what's your big secret to betting correctly?" The mathematician replies, "Oh, yeah, I devised this amazing formula for predicting the winner. I've been testing it out recently, and noticed it seemed to do very well." Intrigued about his methodology, his friends poked him a bit more. "So, how do you do it?" The mathematician then starts out explaining, "Okay, so first thing, I assumed all horses were identical and spherical..."
sophie
Break My Heart
Frankly, I don't see the purpose of this thread in the first place. It is entitled "I don't care what you say" however, by putting this on the forums, you are expecting some sort of responses. So, you post this information, but will not accept any counter-arguments...so why couldn't you just keep your beliefs to yourself and avoided a whole lot of pointless arguing?
Sorry, rant over.
Sorry, rant over.
Indeed. The non-sentence "more math less name-calling" is IMO rude in that it is immediately accusing and not constructive in any way. It's basically passing judgement. But why do you argue semantics?
Isn't it ironic, that of us two I'm the only one who has used math in his posts so far (to refute the methods of the OP), and that even if I was name-calling (it comes down to arguing a definition), you are doing the same thing by slandering me with a name-calling accusation? You take the word I used out of its original context and drag me into a pointless and off-topic argument. I really don't understand why you felt the need to do so outside of PM'ing me.
SillyGoat
Warlord
Bigv32: nice troll. 
Jet
No, no, please. Please.
- Joined
- Mar 16, 2006
- Messages
- 2,431
Your criticism of the ethics or style of my criticism doesn't do anything to defend what you said in the first place.
Post #18 does not require one to take your word out of context. The context you referred to in post #40 - defective statements made by the original poster - doesn't make your use of the word "laughably" in post #17 any less derogatory, bullying, or mathematically inadequate.
Post #18 does not require one to take your word out of context. The context you referred to in post #40 - defective statements made by the original poster - doesn't make your use of the word "laughably" in post #17 any less derogatory, bullying, or mathematically inadequate.
I will not be drawn in, Jet. I have nothing else to say to you.
PieceOfMind said:
Let me preface this by saying I'm drunk and not entirely sure why I feel the need to enter this argument and:
1) Someone has started a thread stating they believe the combat odds generator to be mathematically incorrect and
2) failed to post anything to sufficiently back-up their claims that they believe this to be true and
3) I realize that I'm a new poster so I apologize to anyone who takes offense to this post (except the person I'm patronizing). AND
4) Don't feed the trolls
Really?
Let's continue derailing topics. All you've done so far is complain that he's used an adjective to describe something and called it name-calling. Let's see you post some proof that what he's doing is actually name calling. (since you're calling him out on posting insufficient proof of his responses)
de⋅rog⋅a⋅to⋅ry
1. tending to lessen the merit or reputation of a person or thing; disparaging; depreciatory: a derogatory remark.
Wikipedia on name-calling:
When employed, name-calling is thus a substitute for rational, fact-based arguments against an idea or belief, based upon its own merits.
http://en.wikipedia.org/wiki/Name-calling
***These two definitions show how derogatory remarks and name calling are independent from one another.***
I personally fail to see how PieceOfMind calling 100 trials laughably insufficient is name-calling when he backs his statement up with:
So whether or not you agree with his reasoning of 100 trials being insufficient he offers "proof" on why he believes 100 trials to be insufficient information - regardless of the truth of such a statement - thus negating the definition of name calling.
You poopy-headed ignant bastid.
^ name calling.
Additionally while he could have supported his remark with math as Peppe1 did by employing the use of google considering he is the author of the Advanced Combat Odds calculator seems as though it would be overkill, especially considering that the title of the post was: "I do not care what you say"
1) Someone has started a thread stating they believe the combat odds generator to be mathematically incorrect and
2) failed to post anything to sufficiently back-up their claims that they believe this to be true and
3) I realize that I'm a new poster so I apologize to anyone who takes offense to this post (except the person I'm patronizing). AND
4) Don't feed the trolls
Really?
Let's continue derailing topics. All you've done so far is complain that he's used an adjective to describe something and called it name-calling. Let's see you post some proof that what he's doing is actually name calling. (since you're calling him out on posting insufficient proof of his responses)
de⋅rog⋅a⋅to⋅ry
1. tending to lessen the merit or reputation of a person or thing; disparaging; depreciatory: a derogatory remark.
Wikipedia on name-calling:
When employed, name-calling is thus a substitute for rational, fact-based arguments against an idea or belief, based upon its own merits.
http://en.wikipedia.org/wiki/Name-calling
***These two definitions show how derogatory remarks and name calling are independent from one another.***
I personally fail to see how PieceOfMind calling 100 trials laughably insufficient is name-calling when he backs his statement up with:
PieceOfMind said:
So whether or not you agree with his reasoning of 100 trials being insufficient he offers "proof" on why he believes 100 trials to be insufficient information - regardless of the truth of such a statement - thus negating the definition of name calling.
You poopy-headed ignant bastid.
^ name calling.
Additionally while he could have supported his remark with math as Peppe1 did by employing the use of google considering he is the author of the Advanced Combat Odds calculator seems as though it would be overkill, especially considering that the title of the post was: "I do not care what you say"
sophie
Break My Heart
*Sigh* This thread is getting rather pointless...not to mention completely off-topic. Can we please lock it and move on?
Getting 38 wins from 100 battles gives a chi-square of 5.76, if I calculated it correctly. Thus the probability that the null hypothesis (i.e. the hypothesis that the RNG is working correctly) can explain the result correctly is about 0.0163. Most statisticians appear to argue that "If the probability that the null hypothesis can explain the experimental results is above 1%, an experiment is generally not considered evidence of a different hypothesis". (source: http://www.fourmilab.ch/rpkp/experiments/statistics.html)
Therefore winning 38 battles from 100 is insufficient to argue for a hypothesis other than the RNG working correctly.
Unfortunately for the OP, he continued the experiment by doing a further 100 tests which only moved the results back into the more reasonable range.
Therefore winning 38 battles from 100 is insufficient to argue for a hypothesis other than the RNG working correctly.
Unfortunately for the OP, he continued the experiment by doing a further 100 tests which only moved the results back into the more reasonable range.
JBossch
King
This is definitely the best state in which to weigh on this BS thread.
It should also be noted that in fact the RNG is not being called to produce a random binary variable here for each battle. The RNG would be used to realize a random variable several times during each battle. In theory this should not change the probabilities of the more macroscopic outcomes we observe (victory or death from battles) but since it is merely a pseudo-RNG it is quite possible it is more vulnerable to tests like the ones we're conducting.
My guess would be that this would usually dampen the streaky behaviour of battle outcomes.
It's interesting the RNG is less often called into question in the other parts of the game where it is used (e.g. GP births and holy city spawns) than it is for battles. This could be because battles are far more common and are subject to greater scrutiny or it could be because we are more biased to expect the favourable result from battles, since it is costing us when the RNG spits the less expected string of values.
My guess would be that this would usually dampen the streaky behaviour of battle outcomes.
It's interesting the RNG is less often called into question in the other parts of the game where it is used (e.g. GP births and holy city spawns) than it is for battles. This could be because battles are far more common and are subject to greater scrutiny or it could be because we are more biased to expect the favourable result from battles, since it is costing us when the RNG spits the less expected string of values.
Hmm all I can think of when reading the first page of this thread is: 'who cares?', seems a right waste of time arguing over something as frivolous as this. Even if the odds were slightly off, it adds a random element to the game, making it more fun. 
Dirk1302
Deity
Instead of doing tedious manual battles and not checking the combat odds in the combat log it'd be easier to test the rng in the sdk where it's easy to test the thing ~1000000 times. Not going to do it however since i'm confident it's correct.
Similar threads
