[Game] Umineko no Naku Koro ni Chiru - Ep. 5 End of the golden witch

[Jan-Poo]

As I said in my premise this is the real world logic.

If your claim is that Erika thinks as if she is inside a mystery novel I certainly won't disagree. However this kind of reasoning is far from being logic, unless Erika really was inside a mystery novel.

I'll be clear. What I am actually arguing is that there are only two possibilities:

1) Erika is going against the knox rules by the fact even so she's using wrong logic she happens to find the right murder spot. Lucky.

2) Erika is inside a novel where novel rules apply therefore her reasoning is correct.

of course I choose number 2

[Klashikari]

Originally Posted by Jan-Poo

http://en.wikipedia.org/wiki/Murder

search for japan.

In Japan there is only 1 murder over 200.000 persons per year. Since the family meetings happens only once time per year I can agree that the 4 october represents this 1 per year chance of a murder occurring.

this basically means that on a randomly selected sample of 100 japanese people you'd have a 0.0005% of probability to get one that committed a murder that year.

On 18 people that would become 0.00009%

Again, why are you applying statistics on humans? And to "global" setup to boot?
Again, stastics cannot work over a specific case, they are representative as a summary to a given subject, to a given population. By no mean it can be applied to everyone on a smaller sample. Moreover, this kind of sample doesn't mean jack when the given situation isn't even represented to the whole country.

That's like suddenly saying that "no matter what, you will have 50% chance of having head, 50% having tails by flipping a coin", and yet, there is nothing that prevents you to have a odd result, having 8 heads, and 22 tails after 30 coin flips.
We can't apply probabilities and statistics to human beings, let alone when you have peculiar circumstances in a given situation and possible backstory etc.

Realistic or not, you really cannot use an observed statistics to be possibly applied as if these 18 persons were in the same situation (or remotely the same) of a whole population.

[Antera Caramichael]

Originally Posted by Jan-Poo

As I said in my premise this is the real world logic.

If your claim is that Erika thinks as if she is inside a mystery novel I certainly won't disagree. However this kind of reasoning is far from being logic, unless Erika really was inside a mystery novel.

And she is, look a her beahavior, at how she is quoting mystery novels, how she calls herself "intellectual rapist", how she is searching for riddles, trying to create even some.
in a way, she make me think of Maria, who can quote you any sentence of the Bible in hebrew. Erika is really light comparing to her...

[rogerpepitone]

Klashikari: Your foreshadowing hints seem to me to be the equivalent of "The detective suddenly stares at the ground, says "Aha", and picks an unknown object up out of some corner. Ten chapters later, during the big reveal, he produces the object, explains what it is, and how it incriminates the suspect."

The whole point of the Knox rules is that the reader is supposed be on an even footing with the detective, and should be able to perform the same deductions. It's not enough to know that some passages were sealed in a way that they couldn't be used; the reader has to know which passages were sealed and when.

[Klashikari]

Originally Posted by rogerpepitone

Klashikari: Your foreshadowing hints seem to me to be the equivalent of "The detective suddenly stares at the ground, says "Aha", and picks an unknown object up out of some corner. Ten chapters later, during the big reveal, he produces the object, explains what it is, and how it incriminates the suspect."

The idea is that the clues were setup.
I didn't claim they were reastically done. To begin with, we can all agree that Erika was blatantly setting up, with all the crappy joust in the meta world.

The whole point of the Knox rules is that the reader is supposed be on an even footing with the detective, and should be able to perform the same deductions. It's not enough to know that some passages were sealed in a way that they couldn't be used; the reader has to know which passages were sealed and when.

That work for the original application of the knox rule, but in umineko, the application of the rules are sort of different. There isn't nothing about the "readers" here, but how the rules are followed by the characters within their own scope of actions. Otherwise, there wouldn't be any weird rewording of the said rules, especially the ninth one.

The reasoning behind the sealing was already introduced by Eva a while ago, and we were presented Erika meddling around. That said, that "clue" isn't exactly self explanatory either.

[Jan-Poo]

Originally Posted by Klashikari

Again, why are you applying statistics on humans? And to "global" setup to boot?
Again, stastics cannot work over a specific case, they are representative as a summary to a given subject, to a given population. By no mean it can be applied to everyone on a smaller sample. Moreover, this kind of sample doesn't mean jack when the given situation isn't even represented to the whole country.

That's like suddenly saying that "no matter what, you will have 50% chance of having head, 50% having tails by flipping a coin", and yet, there is nothing that prevents you to have a odd result, having 8 heads, and 22 tails after 30 coin flips.
We can't apply probabilities and statistics to human beings, let alone when you have peculiar circumstances in a given situation and possible backstory etc.

Realistic or not, you really cannot use an observed statistics to be possibly applied as if these 18 persons were in the same situation (or remotely the same) of a whole population.

Nice example you made there but you are totally misinterpreting me.

I am not saying that it is impossible for a coin toss to result in head or tail. I'm saying that if you guess it right you are lucky.

Not you don't need to be very lucky if you guess right that a coin toss will result into a had. that's a 50% probability

However if you guess right something that has a 0.00009% probability to happen, then how is that not supposed to be lucky?

Oh and statistic on humans work. You are going against 50 years of social psychology by disagreeing with this!

[Judoh]

Originally Posted by Jan-Poo

http://en.wikipedia.org/wiki/Murder

search for japan.

In Japan there is only 1 murder over 200.000 persons per year. Since the family meetings happens only once time per year I can agree that the 4 october represents this 1 per year chance of a murder occurring.

this basically means that on a randomly selected sample of 100 japanese people you'd have a 0.0005% of probability to get one that committed a murder that year.

On 18 people that would become 0.00009%

this is how much probable this would be in a realistic case scenario.

I don't think real world statistics are really "rules": that can be applied to Umineko even if your not following the fiction theory. However I think if your going to use statistics you should use statistics from 1986. I mean If your going with Japan's modern day statistics it would be a 90% probability that everyone committed suicide. That is... if you didn't take the red truth into account...

Also even if the chance is low it's not impossible. In Ryukishi's anti fantasy vs Antimystery article he called something that will certainly happen a 99.99% chance and a miracle something that has a .01% chance to happen.

[Jan-Poo]

You are missing my point, read above.

[bereal31]

anyways to me...those clues are invalid since they are not revealed till the end of the mystery. all the forshadowing that you mentioned are TOO GENERAL to mean anything.

she was late. so? she could've tripped. had another cup of tea. took a leak. if she was late and had a piece of duct tape on her cloth or carrying a duct tape though, for example, that would make it a valid foreshadowing for this clue.

she was meddling around. so? all girls her age are like that. ever had an annoying, nosy sister? girls don't just all of suddenly become wives, you know. if she was meddling around and was found crouched down in front of one of those doors, for example, that would be valid.

see, battler saying at the end that he was the boy from 19 years ago is valid because the forshadowings were SPECIFIC and pointed toward that direction.

for example: his dad wanting to talk about his birth/his origins.

[maximilianjenus]

the foreshadowing works for me, we had alread ot the paper on kinzo's door back in Ep1, this Ep now gives a more valid reason on why eva did that ( to corner natsuhi into accepting kinzo's already dead), plus we got genji's door sealed with tape, all that happened long before the trial.

now, to maek the probability of erika guessing the murder room less ridiculous think of it differently, analyze the way the seals were set up adn where, now make the crime happen in another room.

Let's say everybody outside the cousin's room got killed, if natushi survived and her room was hte only one with a broken seal besides the ones in which murders were commited we got a clear culprit too.

[Jan-Poo]

Erika knew better than place a seal on Natsuhi's door. If she did that, she'd give Natsuhi a perfect alibi. No, she never did that. Natsuhi's door wasn't sealed.

Also as Erika stated she expected 6 people to die no more than that.

It was possible that the murder would happen to anyone that wasn't ins the guesthouse. It was even possible that Battler and Rosa would be murdered there instead than in the Guesthouse... hell it never happened before that someone died in the guesthouse I don't think I'm saying something far-fetched here.

[ijriims]

Let me place myself in Erika's shoe:

So I have come to an isolated island with a rich family arguing about inheritance. And there was an epitaph shown by the family head as well. Based on these and my experience that everywhere I go there must be a murder, the occurance of a murder on Rokkenjima has a probability close to 1.

I am a detective, not a police, so my job is not to safeguard everyone but to solve a mystery. And really why do I have to prevent a murder at all even if I know it is coming? My only enjoyment is to find out the culprit and show it off to everyone else.

Now the question is: when will the murder happen, who will be killed and where will it be? Now I have successfully solved the epitaph and presented it to everyone. And the adults were getting into hot quarral again, Fu..Fu..Fufu... This must have triggered the murder now.

With Battler being the new head of the family and thus the centre of attention, he had the highest probability of being killed of all people. Of course, the ones who opposed Battler as the next family head like Krauss and Natsuhi would also have a high chance of being killed by the other adults. But anyway, now they were all in the mansion right now and probably the murder would not happen under the sight of other people, it must be done in secret.

Yes, probably when people were being defenseless, such as during sleeping. Isn't it that most people are going to rest in the guesthouse? Yes, the murder should happen inside the guesthouse and at midnight while Battler was the most probable victim. I should be stationing in the ground floor of the guesthouse to monitor who enter or leave the guesthouse. At the same time, since I cannot monitor the 2/F while I am on the ground floor, I would seal up all windows on the second floor so that no one could slip into the second floor without my noticing it. Nanjo's room and Kumasawa's room doors should be sealed up as well so if they were the murderers, the seal would be broken and I can quickly pinpoint them.

There is also a good chance of Krauss and Natsuhi being murdered. Well, i cannot be in the mansion at the same time. Still, they would not be killed when they are hotly debating with other adults. Probably they will be killed inside their rooms, when they are asleep. In order to pass through their door, maybe the master keys would be involved. With Genji being locked inside the waiting room by Eva, Kumasawa being locked in the guesthouse, Gohda being locked inside guesthouse's servants'room after I have been talking with him, the only servants left would be Shannon and Kanon. So if Krauss and Natsuhi were murdered, they should be involved.

Or maybe a murder like" Murder on the Orient Express" will happen, and the other adults kill Krauss and Natsuhi right inside the parlor and collude with each other. That will be INTERESTING. Will Eva actaully lying about helping me? It will be fun to find out. But the collusion of all other adults is less likely than Krauss and Natsuhi being killed in secret, and much less than Battler being killed.

Or maybe all adults except Krauss and Natsuhi will be killed. Krauss and Natsuhi definitely have the motive to eliminate other adults, who are supporting Battler's as the next head. OF course, it is much more complicated and risky than just murder Battler. Would they kill all other adults in one shot like poisoning drinks and food in the parlor? Well, this is unlikely as this would put too much suspicion on them. So even if Krauss and Natsuhi wanted to murder other adults, they needed to do it separately.
-----------------------

I can go on and justify Erika's actions with logics and realistic setting, provided "everywhere she go, murders happen" is true. One could assign probability of being victim for every person on Rokkenjima through logics and then set up a plan on how to setup evidence of alibi for some people while not preventing the occurance of a murder.

[Klashikari]

Originally Posted by Jan-Poo

Nice example you made there but you are totally misinterpreting me.
I am not saying that it is impossible for a coin toss to result in head or tail. I'm saying that if you guess it right you are lucky.

Not you don't need to be very lucky if you guess right that a coin toss will result into a had. that's a 50% probability

However if you guess right something that has a 0.00009% probability to happen, then how is that not supposed to be lucky?

You are using stastitics to elaborate a probability. By itself, I really don't get how you can join both concepts, considering the obvious shades of greys and circumstances regarding that kind of deed.
By the way, my example shows that even probability can be totally off tracks (the usual Gambler's fallacy).

Oh and statistic on humans work. You are going against 50 years of social psychology by disagreeing with this!

Although statistics work on humans, what you are doing is totally misplaced completely.
You are comparing statistics of a whole country to a very small sample of individuals. By this point, there is absolutely no way you can claim that these 18 are enough to represent that much of sample of this statistics. Otherwise, are you claiming that the 18 people you are picking randomly will represent all type of classes, age, gender etc? I beg to differ.

Again, my point is that statistics doesn't work on such low sample, especially because of circumstances. Statistics work only on a certain target population about a subject, when they are defined to a certain point, but also if they are numerous enough. The statistics you presented there represent a whole frigging country that has its share of different classes of people, with all circumstances you can imagine.
Now, if I were, for example, to select 20 people from a ghetto, or 20 people from a prison, do you think the probability will remain the same as to the statistic of a whole country? I really doubt so. These do not represent the country at all, but the statistics I would gather in such extreme example would be extremely high. Now, if you merge these individuals with more "normal" people in the lot, the statistics would be affected greatly. Likewise, if you keep your observation on a very specific kind of individuals, you will have a very different result.

Thus, you can't probe that probability with statistic of a unrelated observation that is just reporting a number affecting a country, and not a certain type of population.
We are talking here of a probability of a murder occuring with bunch of people in DIRE need of money, barking at each other for decades, with a very elusive family head, an unhealthy atmosphere of occult, a typhoon trapping everyone, and suddenly, a (un)fortunate youth having all of a sudden 20 billion yens, that everyone would turn berserk over it, in his pocket.

Side note: this note doesn't work on the premise of the scheme of the meta world. In a realistic setup though, the following explanation can apply:
Erika didn't expect 6 murders to begin with (if we don't account the meta world, that is). She knew a murder will occur, and the victim would be Battler. However, AFTER they discovered there were 4 casualties, she then says there might be others, which Kyrie agreed.
None claimed there would be 6 victims, but as soon as they knew there weren't only 1, the conclusion was easy.

[Jan-Poo]

@ijriims

If you really took your time to read my posts you'd have noticed that I have already demonstrated the fallacy of this argument. Yes Erika thought Battler was going to murdered. But Battler wasn't murdered at all, he wasn't even the target.

I'll repeat it for the last time:

If you get to the right spot because of a right reasoning it's your merit.
If you get to the right spot because of a completely wrong reasoning, you have no merit whatsoever you have just been lucky.


@Klash
This whole discussion on statistic is going kinda overboard, I'm not going to get on details because this isn't really the place for this but you are making a lot of assumptions that are completely wrong, for example one of the very point of statistic is to generalize data from a small sample to a larger population and here you are saying that those statistical data are about a whole frigging country and can't apply to a small population? I think you are getting it all wrong.

Also you are probably completely ignoring what I said earlier. What I am contesting here is the concept in a general sense that it is "obvious that people would kill for 20 tons of gold". And I have precisely said that there are more margins of discussion about "It was obvious that Natsuhi or Krauss would kill Battler considering the situation". There is a whole world of difference between such statements.

For example you talk about how they were in dire need of money. This talks about a specific situation and totally excludes the servants from the picture, this would be a good point to raise the probability for this particular sample to commit murder... except they aren't really in dire need of 20 tons of gold, and the money they are getting anyway are a lot more than what they hoped to get. Sorry this point is completely moot.

Frankly the only thing that you can say is that according to the situation there were signs that Krauss or Natsuhi would kill Battler, there was absolutely no sign nor any reason from anyone else that they were desperate enough to kill in any way. All the others were quite happy about the whole situation. No one was in "dire needs" to have Battler succeed the headship and get half of the gold, expecially not Eva Hideyoshi and Rosa. And at any rate Rudolf and Kyrie wouldn't have any reason to kill Battler nor anyone inside the cousin's room.

[MeoTwister5]

@Jan-Poo

What Klash is pointing out is that the reason you can't apply the same inferences and conclusions from a larger population to the discrete number of individuals on the island is because the idea of sampling is randomization such that you are assuming that the samples you chose are randomly taken from the population at large such that they are representaive of the general qualities the population possesses to make your conclusions "valid" for the population from which they from.

The problem is that the the Rokkenjima people are not scientifically representative of the population they are culled from. Discrete random sampling compared to whole population studies can only be concretely related insofar as all the variables for them are are the same; randomization assumes that the sample qualities and variables are the same as the population and therefore representative. Heck even then, comparative analyses such as regression, chi square, T tests and the like have to be applied to even validate such conclusions

Your logic that a large population's satistical data must always apply to a sample culled within it is flawed because there is ALWAYS something that causes a deviation. A 100% overlap is realistically impossible. Even the most intricate mathematical models for randomization cannot create a 100% coincidence. The fact that the 17/18 people of Rokkenjima have far too many similarities, confounders and the like already puts any population models out the window. They are not representative of the populations they are in, and therefore the data of the general population cannot accurately apply to them.

Heck if someone managed to turn "murder qualities" of the Rokkenjima people into statistical data and do a T-Test comparison to a randomized sample of the general Japanese people, I can guarantee you it will be nowhere even near a 95% confidence interval.

[rogerpepitone]

Except Erika's actions seem to be chosen with the purpose of tying down everyone but Natsuhi, while leaving Natsuhi completely free. (Such as having Genji, who could reasonably be called out at any moment, sealed in, but not having Krauss's or Natsuhi's rooms sealed.)
And most of them wouldn't have been any use against a slightly different crime. (If the seal on one of the guestroom windows had been broken, it would have just opened up everything.)

And when they're trying to get into Kinzo's room the next day through the window, she doesn't mention any ladder or offer to show off her freeclimbing skills.

[maximilianjenus]

Originally Posted by Jan-Poo

Erika knew better than place a seal on Natsuhi's door. If she did that, she'd give Natsuhi a perfect alibi. No, she never did that. Natsuhi's door wasn't sealed.

I might need to read it again, but I think natsuhi's seal was "broken" just like a few other seals which is why erika was setting up guard.

[Jan-Poo]

Originally Posted by MeoTwister5

@Jan-Poo

What Klash is pointing out is that the reason you can't apply the same inferences and conclusions from a larger population to the discrete number of individuals on the island is because the idea of sampling is randomization such that you are assuming that the samples you chose are randomly taken from the population at large such that they are representaive of the general qualities the population possesses to make your conclusions "valid" for the population from which they from.

The problem is that the the Rokkenjima people are not scientifically representative of the population they are culled from. Discrete random sampling compared to whole population studies can only be concretely related insofar as all the variables for them are are the same; randomization assumes that the sample qualities and variables are the same as the population and therefore representative. Heck even then, comparative analyses such as regression, chi square, T tests and the like have to be applied to even validate such conclusions

Your logic that a large population's satistical data must always apply to a sample culled within it is flawed because there is ALWAYS something that causes a deviation. A 100% overlap is realistically impossible. Even the most intricate mathematical models for randomization cannot create a 100% coincidence. The fact that the 17/18 people of Rokkenjima have far too many similarities, confounders and the like already puts any population models out the window. They are not representative of the populations they are in, and therefore the data of the general population cannot accurately apply to them.

Heck if someone managed to turn "murder qualities" of the Rokkenjima people into statistical data and do a T-Test comparison to a randomized sample of the general Japanese people, I can guarantee you it will be nowhere even near a 95% confidence interval.

However what I am pointing out that the statement "it is obvious that people would kill for 20 tons of gold" is false. And that is per se a general statement that doesn't include in the picture any of the peculiar factors you are mentioning.

It is also perfectly correct to apply general statistic on people you know nothing about, of course if you add other elements to the pciture the probability changes accordingly. However I again repeat that my use of statistic was finalized to debunk that general statement, and therefore I ask you to explain where in that regard you can find any wrong in my logic.

[blitz1/2]

Originally Posted by ijriims

Let me place myself in Erika's shoe:

So I have come to an isolated island with a rich family arguing about inheritance. And there was an epitaph shown by the family head as well. Based on these and my experience that everywhere I go there must be a murder, the occurance of a murder on Rokkenjima has a probability close to 1.

I am a detective, not a police, so my job is not to safeguard everyone but to solve a mystery. And really why do I have to prevent a murder at all even if I know it is coming? My only enjoyment is to find out the culprit and show it off to everyone else.

Now the question is: when will the murder happen, who will be killed and where will it be? Now I have successfully solved the epitaph and presented it to everyone. And the adults were getting into hot quarral again, Fu..Fu..Fufu... This must have triggered the murder now.

With Battler being the new head of the family and thus the centre of attention, he had the highest probability of being killed of all people. Of course, the ones who opposed Battler as the next family head like Krauss and Natsuhi would also have a high chance of being killed by the other adults. But anyway, now they were all in the mansion right now and probably the murder would not happen under the sight of other people, it must be done in secret.

Yes, probably when people were being defenseless, such as during sleeping. Isn't it that most people are going to rest in the guesthouse? Yes, the murder should happen inside the guesthouse and at midnight while Battler was the most probable victim. I should be stationing in the ground floor of the guesthouse to monitor who enter or leave the guesthouse. At the same time, since I cannot monitor the 2/F while I am on the ground floor, I would seal up all windows on the second floor so that no one could slip into the second floor without my noticing it. Nanjo's room and Kumasawa's room doors should be sealed up as well so if they were the murderers, the seal would be broken and I can quickly pinpoint them.

There is also a good chance of Krauss and Natsuhi being murdered. Well, i cannot be in the mansion at the same time. Still, they would not be killed when they are hotly debating with other adults. Probably they will be killed inside their rooms, when they are asleep. In order to pass through their door, maybe the master keys would be involved. With Genji being locked inside the waiting room by Eva, Kumasawa being locked in the guesthouse, Gohda being locked inside guesthouse's servants'room after I have been talking with him, the only servants left would be Shannon and Kanon. So if Krauss and Natsuhi were murdered, they should be involved.

Or maybe a murder like" Murder on the Orient Express" will happen, and the other adults kill Krauss and Natsuhi right inside the parlor and collude with each other. That will be INTERESTING. Will Eva actaully lying about helping me? It will be fun to find out. But the collusion of all other adults is less likely than Krauss and Natsuhi being killed in secret, and much less than Battler being killed.

Or maybe all adults except Krauss and Natsuhi will be killed. Krauss and Natsuhi definitely have the motive to eliminate other adults, who are supporting Battler's as the next head. OF course, it is much more complicated and risky than just murder Battler. Would they kill all other adults in one shot like poisoning drinks and food in the parlor? Well, this is unlikely as this would put too much suspicion on them. So even if Krauss and Natsuhi wanted to murder other adults, they needed to do it separately.
-----------------------

I can go on and justify Erika's actions with logics and realistic setting, provided "everywhere she go, murders happen" is true. One could assign probability of being victim for every person on Rokkenjima through logics and then set up a plan on how to setup evidence of alibi for some people while not preventing the occurance of a murder.

U know, this makes a great bonus TIP. (maybe Ryukishi should have written Erika's letter/diary or something.

[MeoTwister5]

Originally Posted by Jan-Poo

However what I am pointing out that the statement "it is obvious that people would kill for 20 tons of gold" is false. And that is per se a general statement that doesn't include in the picture any of the peculiar factors you are mentioning.

It is also perfectly correct to apply general statistic on people you know nothing about, of course if you add other elements to the pciture the probability changes accordingly. However I again repeat that my use of statistic was finalized to debunk that general statement, and therefore I ask you to explain where in that regard you can find any wrong in my logic.

What I I find wrong in your logic is that you tried to use generalized statistics to debunk a claim that cannot entirely apply to a compromised sample set, no matter how ludicrous the claim may be. The statement "anyone would obviously kill for 20 billion yen" is a completely ludicrous a priori assumption by almost anyone's standards, but again no matter how ludicrous the statement is the evidence presented does not jive with rebuttal. What if to say that the people on Rokkenjima are so deviated from the norm that "anyone" (referring to everyone on the island) would kill for 20 billion yen? This obviously would not coincide with everyone else.

The thing is we don't know enough about the Rokkenjima people to make a connection from them to them to the rest of Japane, and the inverse the same. Even if the statement is absurd, we can't EXACTLY say that it's a statement that cannot describe them, because until we know such things completely is fairly possible that such a statement can apply to the Rokkenjima people.

Ugh. It's the damn CatBox all over again.

Edit - And I honestly don't want to discuss all this statistics any further than it has to. There's a reason I got a friend to do the statistics on my undergrad thesis.