AnimeSuki Forums

Register Forum Rules FAQ Community Today's Posts Search

Go Back   AnimeSuki Forum > Anime Related Topics > General Anime

Notices

Closed Thread
 
Thread Tools
Old 2011-10-29, 10:39   Link #4141
G.o.S
~desu♥
 
 
Join Date: Jun 2006
Location: Falls Church, VA
Kyoko won a hard match today and step into the semi-final with Hartmann, who definitely is the dark horse of this year. From today result, I think everyone will expect Kyoko to beat her opponent easily to go into the final. If that's the case then I would like to have Sayaka be there with her. This will make a very sweet saimoe final match. Not that I dislike any other remaining contestants. In fact, I like all of them and wouldn't mind seeing any of these girl winning the trophy.
For today, I'll be rooting for Sayaka. It seems like some people predict Madoka will win since she's the main character of her series. I see her as the weaker girl because she didn't make such a huge impact in the anime compared to Homura, Sayaka, Kyoko and even Mami, but we'll see what happen today.
__________________
G.o.S is offline  
Old 2011-10-29, 10:45   Link #4142
FlavorOfLife
Uncaring
 
 
Join Date: Sep 2010
1:00- 1:59 131/131 83/ 83
2:00- 2:59 28/159 19/102
3:00- 3:59 9/168 15/117
4:00- 4:59 14/182 11/128
5:00- 5:59 6/188 2/130
6:00- 6:59 6/194 11/141
7:00- 7:59 9/203 14/155
8:00- 8:59 13/216 15/170
9:00- 9:59 9/225 17/187
10:00-10:59 8/233 9/196
11:00-11:59 4/237 11/207
12:00-12:59 14/251 19/226
13:00-13:59 10/261 14/240
14:00-14:59 8/269 6/246
15:00-15:59 11/280 17/263
16:00-16:59 15/295 17/280
17:00-17:59 20/315 15/295
18:00-18:59 15/330 16/311
19:00-19:59 19/349 27/338
20:00-20:59 21/370 31/369
21:00-21:59 26/396 39/408
22:00-22:59 249/645 150/558
23:00-23:59 12/657 10/568

Looks like an immense absolute gap of 12 votes before the final hour. Kuroneko absolutely widened that gap to 12 votes to where Kyoko had absolutely no hope of catching up and she only won by cheating. Indeed that 150 votes for Kuroneko was also totally legit and free from any implied cheating which only Kyoko could have done. *facepalm*

A look at the graphs of all MS competitions (exept Sayaka) to the prelims shows the creation and growth of a faction whose tactic is final hour voting backed by normal voting of a general audience. They win because of the general audience support which some fail to realise is the result of the way the series ended itself during a disaster and time of uneasiness
FlavorOfLife is offline  
Old 2011-10-29, 10:47   Link #4143
Eisdrache
Part-time misanthrope
 
 
Join Date: Mar 2007
Quote:
Originally Posted by Coldlight View Post
I miss seeing the AnimeSuki Rush votes on 2ch and all the ASCII art that came with it.
Aw man the ASCII arts and AS rushes were easily one of the best parts even if I never participated in any of them

Also I'd have bet money on Kuroneko, I stand corrected. FML
Eisdrache is offline  
Old 2011-10-29, 10:52   Link #4144
Ichuki
┌(^o^)┘モリ!┌(^o^)┘モリ!
 
 
Join Date: Sep 2010
Prediction- Madoka
Ichuki is offline  
Old 2011-10-29, 10:54   Link #4145
ion475
にこにこにー
 
 
Join Date: Aug 2009
Location: 国立音ノ木坂学院
Sigh...there goes my one and only hope...F**k JST 2011...

Well, not really, if Mami or Ika-chan win I'll still be happy. On the other hand, if even Kuroneko can't stop the 4th tier Madoka girl, who can stop Mami?

As for next round - I'm going with Sayaka...I mean, may as well.

Conclusion? Erica for Saimoe 2011 Crown!!!!!
ion475 is offline  
Old 2011-10-29, 11:00   Link #4146
Arkanis
Senior Member
 
 
Join Date: Jun 2008
Quote:
Originally Posted by wontaek View Post
You are still missing 2 things I said earlier.

1. I did point out that these graphs may be possible if someone coordinates the supporters. Someone already pointed out that 2ch's Madoka channel people likely have talked about such thing. Although I think they took a big risk by concentrating only on the last hour instead of last 2 hours, existence of talk invalidates my base assumption, thus I can no longer claim that the result is likely due to massive cheating by an individual or two.
One thing I don't get: Why is it a big risk? In my opinion the risk is bigger when you draw the votes out over a time frame of 2 hours, since then the opponent could react with own votes she wouldn't get otherwise (that's essentiall the whole deal with the fake votes for Kyouko).
Quote:
2. You cannot focus only on the first and last hour peak when analyzing these graphs. Kuroneko vs Kyouko graph shows a fundamental difference from all the other graphs in that Kuroneko's vote tally slope was consistently larger than Kyouko, and Kuroneko not only overtook Kyouko, but was widening the gap before the last hour block. If you look at the 2nd derivative of the (Kyouko's vote total - Kuroneko's vote total ) vs hour graph, the spike near the 1st hour and last hour is just too large to not arouse suspicion.
You can't simply compare the derivatives of the (Kyouko's vote total - Kuroneko's vote total) function to the respective ones from the other matches since they aren't linear. If you scale Kuroneko's vote count then these values will look different despite being regular. I already mentioned the scaling of Kuroneko's function. (I hope I understood your comment regarding the tally slope correctly.)

Just a little meta comment: Even if there are slight differences, it doesn't matter. In fact, I'd be surprised if there weren't any differences at all. I'm not even sure if we aren't streching the possibility of drawing conclusions out of comparisons too far. Each match has different circumstances, too many variables in the background we aren't aware of.

Quote:
Granted that the existence of discussion of coordination is plausible explanation for such a spike, but then, the conclusion you would reach is not something that would please many people : few people outside the 2ch Madoka channel had voted for Kyouko in this match.
Evidence? (And please don't forget the possibility of coordination without a leader.)

Quote:
Now, if that indeed is the case, then we now have problem explaining why we had so many more voters in this match compared to other matches. Is it simply because this match attracted more attention compared to other matches? Sure, that could be so. It still doesn't wash off that unusual bitter taste lingering in those who are not Kyouko supporters.
I still don't see how the little presumable inconsistencies are enough to reject the 0-Hypothesis that there is no cheating involved. In fact I feel a little... offended by these from my point of view a little careless allegations.
Arkanis is offline  
Old 2011-10-29, 11:08   Link #4147
Triple_R
Senior Member
*Author
 
 
Join Date: Jan 2008
Location: Newfoundland, Canada
Age: 42
Send a message via AIM to Triple_R
Congrats to Kyouko! Now a Saimoe Slam 2011 continuation...


Spoiler for More Saimoe Slam commentary from Kyubey Bischoff!:
__________________

Last edited by Triple_R; 2011-10-29 at 11:23.
Triple_R is offline  
Old 2011-10-29, 11:09   Link #4148
broken270
Defying gravity
 
 
Join Date: Jul 2010
...Now, where were those people who were not participating in most of the entire previous matches from before? It seems almost certain that the final will be a PMMM final. With what was shown, if Ruri cannot defeat Kyouko, what are the chances for Ika to defeat Mami? Result is result for now. I guess I should congratulate Kyouko. At least, it was my second favorite PMMM girl (Except for Homura, everyone else is terrible in my taste)

Still unable to vote, but I will side with Sayaka. She should be able to win, unless Madoka was hiding her strength the entire time.
broken270 is offline  
Old 2011-10-29, 11:26   Link #4149
Arkanis
Senior Member
 
 
Join Date: Jun 2008
Sorry for OT, but does someone know what happened to the easymodo archive?
Arkanis is offline  
Old 2011-10-29, 11:45   Link #4150
zato_1one
Senior Member
 
 
Join Date: Apr 2006
That's it. All my girls are gone now. LOL But it's not that I hate Kyouko so I'm still somewhat happy with the result anyway.

Kyouko want to say thank you for those who cheer her on.



Well, I intend to post this pic instead if Kuroneko win.

Spoiler for Black Ranger:
__________________
zato_1one is offline  
Old 2011-10-29, 12:15   Link #4151
wontaek
Senior Member
 
 
Join Date: Oct 2006
Location: Planet Earth
Age: 54
Quote:
Originally Posted by Arkanis View Post
One thing I don't get: Why is it a big risk? In my opinion the risk is bigger when you draw the votes out over a time frame of 2 hours, since then the opponent could react with own votes she wouldn't get otherwise (that's essentiall the whole deal with the fake votes for Kyouko).

You can't simply compare the derivatives of the (Kyouko's vote total - Kuroneko's vote total) function to the respective ones from the other matches since they aren't linear. If you scale Kuroneko's vote count then these values will look different despite being regular. I already mentioned the scaling of Kuroneko's function. (I hope I understood your comment regarding the tally slope correctly.)

Just a little meta comment: Even if there are slight differences, it doesn't matter. In fact, I'd be surprised if there weren't any differences at all. I'm not even sure if we aren't streching the possibility of drawing conclusions out of comparisons too far. Each match has different circumstances, too many variables in the background we aren't aware of.


Evidence? (And please don't forget the possibility of coordination without a leader.)



I still don't see how the little presumable inconsistencies are enough to reject the 0-Hypothesis that there is no cheating involved. In fact I feel a little... offended by these from my point of view a little careless allegations.
1. If you are comfortable with everyone in the channel be capable of getting the voting done within minutes of desired time without much technical difficulties, then sure it is not a big risk. If you assume that there are all kinds of voters trying to follow your lead, you have to make rooms for newbies and other technical difficulties. Therefore, to make sure you can reign in all the votes you get, you should give yourself more than an hour, in case someone had unforeseen problem with code generator. Of course, there might be reasons why you don't have to worry about that, but if there is, then I would be grateful if someone enlighten me about it. I do need to say one more thing when I saw the distribution. From the post FlavorOfLife I see that the last bar isn't strictly for just 1 hour period. So, if the reason for the last hour spike is more due to how the data was aggregated , then the ring leader wasn't limiting the rally time to just the last hour, thus was somewhat doing what i suggested.

2. I am comparing how the derivatives are changing within the match itself. Also, I am making note that in most matches we have seen so far, the voter at any time can easily be equated with random drawing from the same random distribution function that resulted in the final result. Easy way to explain that would be that, even if you take away the 1st and last hour, not only the winner, but ratio of the votes received won't likely change much. If the winner of 02:00 to 22:00 mark wasn't the eventual overall winner, usually you would find that the match to be close throughout. Thus in most matches you won't likely see huge spike in derivative of vote function. In this match, that spike is too big, when compared to any other matches, no matter how you calculate that derivative.

Spoiler for Vote difference vs time graph:

3. I was not talking about little difference. Most winners usually win 14 + voting hour blocks. Check it out for any Saimoe 2012 match. Kyouko won only 7 while losing 15. If we eliminate 1st and last, Kyouko won only 5 while losing 15. This is the main point I am saying it is unusual.

4. Let's look at the numbers. Only 237 votes ( = 36.1%) for Kyouko outside of first 2 and last 2 voting block hours in the our vote-vs-time graphs. In contrast 306 votes ( = 53.9% ) for Kuroneko outside of First 2 and last 2 voting block hours. Depending on how you interpret the word 'few', I guess you can say that is many, but I think this number is strongly indicative of the influence of those who coordinated Kyouko's victory and perhaps also of general support the character enjoys outside 2ch regulars.

5. I don't think these are little inconsistencies. I kept saying question the results, because I don't think we should take things at face value when these questionable numerical quantities exist. I never said someone cheated because I guesstimate that Null Hypothesis's P-value is still greater than 10%, but I also guesstimate that it is less than 25%.

So, I don't know about you but what I was trying to accomplish with these writings were

1. Show how one can peer into the data to differentiate the suspicious case from the case that really isn't suspicious. Using the same method of counting hourly blocks won, you can actually claim that it is likely that Madoka girls won all the previous matches legitimately, and they would likely have won even if we count the votes of those who aren't the '2ch regulars' or avid faction members : in other word, general innocent public.

2. To discover what are the information I am missing to better understand the results. For example, I would not have known about the existence of 'Madoka Channel' in 2ch, nor that the last bars above 23:00 in the http://fc.jpn.org/ast2011/ actually is from 22:00 - 23:59 . I doubt people would have told me of this had I not made my doubt known.

3. finally, graphs are pretty.
Spoiler for vote total vs time graph:
__________________
wontaek is offline  
Old 2011-10-29, 12:21   Link #4152
Ichuki
┌(^o^)┘モリ!┌(^o^)┘モリ!
 
 
Join Date: Sep 2010
I can't tell who is winning for now but Madoka seems to start get loads of fakes because of that I feel like Madoka is losing X_X
Ichuki is offline  
Old 2011-10-29, 12:40   Link #4153
Ithekro
Gamilas Falls
 
 
Join Date: Feb 2008
Location: Republic of California
Age: 46
So we have magical girls, an invader, and an ace witch.

The invader is the one with a current show.
__________________
Dessler Soto, Banzai!
Ithekro is offline  
Old 2011-10-29, 12:58   Link #4154
Arkanis
Senior Member
 
 
Join Date: Jun 2008
Quote:
Originally Posted by wontaek View Post
2. I am comparing how the derivatives are changing within the match itself. Also, I am making note that in most matches we have seen so far, the voter at any time can easily be equated with random drawing from the same random distribution function that resulted in the final result. Easy way to explain that would be that, even if you take away the 1st and last hour, not only the winner, but ratio of the votes received won't likely change much. If the winner of 02:00 to 22:00 mark wasn't the eventual overall winner, usually you would find that the match to be close throughout. Thus in most matches you won't likely see huge spike in derivative of vote function. In this match, that spike is too big, when compared to any other matches, no matter how you calculate that derivative.

Spoiler for Vote difference vs time graph:
Even under the assumption that everything is random (so no tactical voting in the last (two) hour(s), I suspect that you are making the error of multiple testing here. If we draw peas and check for a correlation between the color and the risk of getting cancer after eating it, then eventually we will encounter such a correlation even if there is none. If we draw matches and check if there is an inconsistency then eventuall we will find one, maybe at some certain Kyouko Kuroneko match.

Quote:
3. I was not talking about little difference. Most winners usually win 14 + voting hour blocks. Check it out for any Saimoe 2012 match. Kyouko won only 7 while losing 15. If we eliminate 1st and last, Kyouko won only 5 while losing 15. This is the main point I am saying it is unusual.
And I think this is a illegitime comparison, because of the circumstances of this match. Let's say Kuroneko has just 0,75 of her current votes, then everything seem's fine. Now add random anti-madoka-fans who cumulate at the end of the tournament and distribute them like the other Kuroneko voters then suddenly they'll exceed Kyouko's votes in the night hours. But that only means two things. The first is that Kyouko's curve has more voters at start and end while Kuroneko's is more balanced and second that Kuroneko is a particularly strong contender compared to all the others so far.

Quote:
5. I don't think these are little inconsistencies. I kept saying question the results, because I don't think we should take things at face value when these questionable numerical quantities exist. I never said someone cheated because I guesstimate that Null Hypothesis's P-value is still greater than 10%, but I also guesstimate that it is less than 25%.

So, I don't know about you but what I was trying to accomplish with these writings were

1. Show how one can peer into the data to differentiate the suspicious case from the case that really isn't suspicious. Using the same method of counting hourly blocks won, you can actually claim that it is likely that Madoka girls won all the previous matches legitimately, and they would likely have won even if we count the votes of those who aren't the '2ch regulars' or avid faction members : in other word, general innocent public.

2. To discover what are the information I am missing to better understand the results. For example, I would not have known about the existence of 'Madoka Channel' in 2ch, nor that the last bars above 23:00 in the http://fc.jpn.org/ast2011/ actually is from 22:00 - 23:59 . I doubt people would have told me of this had I not made my doubt known.

3. finally, graphs are pretty.
Spoiler for vote total vs time graph:

My problem with this thing is not your desire to analyze things but the following: Nobody questioned the legitimacy of Misaka's wins even though she was weak in comparison at the last tournaments. Nobody even said that there could be cheating at work, even though such numbers should raise suspicion. The thing is, that you must be rather sure that you don't see just things, because otherwise you are spoiling the fun for other people even by just implying that Kyouko's win could result from cheating. Yes, sure, everything could be manipulated, but as long as you don't have very, and I mean very strong arguments, you shouldn't try to make these things a big topic.
Arkanis is offline  
Old 2011-10-29, 12:58   Link #4155
Triple_R
Senior Member
*Author
 
 
Join Date: Jan 2008
Location: Newfoundland, Canada
Age: 42
Send a message via AIM to Triple_R
Quote:
Originally Posted by Ichuki View Post
I can't tell who is winning for now but Madoka seems to start get loads of fakes because of that I feel like Madoka is losing X_X
First hour votes are usually legit, or at least mostly legit.

I hate to say it, but this is probably it for Sayaka. I have to think that this massive 1st hour lead for Madoka is indicative of a solid majority of PMMM faction voters preferring Madoka over Sayaka. I think that Sayaka has some of her own voters (aside from the PMMM faction voters) since she topped 500 twice (whereas Madoka never did, at least not yet) but I don't think it'll be enough to overcome that huge early lead for Madoka.

Ah well, at least Sayaka made it to the Final Eight, if nothing else.

Also, I don't see why people would be cheating in an all-PMMM match-up when the biggest threats to an eventual PMMM Saimoe winner are likely all gone (Biri Biri, Kanade, Kuroneko).
__________________
Triple_R is offline  
Old 2011-10-29, 13:02   Link #4156
Arkanis
Senior Member
 
 
Join Date: Jun 2008
Quote:
Originally Posted by Ichuki View Post
I can't tell who is winning for now but Madoka seems to start get loads of fakes because of that I feel like Madoka is losing X_X
Oh god, erm, I mean.... Oh Madoka, please let it happen. Sacrifice yourself for the Saiyuri Pair 2011! Oh no, these numbers, 132:55, no!
Arkanis is offline  
Old 2011-10-29, 13:15   Link #4157
Micchi
兄さんが望むなら...
 
 
Join Date: Aug 2007
Location: Ottawa, Canada
Pick: 0-1 (0-2, 84-70)
Prediction: 1-0 (1-0, 118-36)

Nice try, Ruri-chan. Time to go read v9.

Pick: Madoka
Prediction: Madoka

If a Madoka char is to win the tourney, might as well be the one that the series is named after.
__________________
Micchi is offline  
Old 2011-10-29, 13:26   Link #4158
Coldlight
Sayaka★Magica
 
 
Join Date: Sep 2010
Location: Under the piercing blue sky
*tightens headband* This could very well be my last stand. No surrender, no regrets.

Spoiler for my vote:

Spoiler for facing mortal danger without help from anyone but herself:


Pick:
QF3: Sayaka Miki - In your moment of greatest need, I stand by you, until the end.

いつだって変わらない
It won't ever change;

思いを残すの。
these emotions which have been left behind.
__________________
Coldlight is offline  
Old 2011-10-29, 13:28   Link #4159
anisu
Member
 
Join Date: Jun 2011
Quote:
in case someone had unforeseen problem with code generator
People would be reserving codes in advance. The "missing votes" from the middle hourly bars would mostly go through the code generator at the normal times, and only the posting on 2ch had to be delayed.

The risk of voters not listening to instructions or 2ch going down is real, but has to be balanced against the benefit of having an "end of match party" to encourage the casual fans to participate. Most of them won't wait around for 2 hours after voting, but half an hour isn't so bad, so voting becomes a more exciting event. The tradeoff probably depends on what kind of fanbase you're dealing with.

Also, regarding "few" votes, Kyouko was approaching 400 votes before the last hour, which is almost as high as her final total in the group final against Index and more than Erica won with in the other quarterfinal. If a number larger than the last quarterfinal's winning total looks like "few", that really says something about how much of a worthy opponent Kuroneko was.

I think it is a good idea to closely check all matches to make sure everything's legit, but after looking at the voting thread I find this one believable.

Quote:
Nobody questioned the legitimacy of Misaka's wins even though she was weak in comparison at the last tournaments.
She did have that very close match against Azusa last time. That's probably a big reason why there wasn't questioning.
__________________
http://ani-list.blogspot.com/ Saimoe Analysis: Serious ------- Business
anisu is offline  
Old 2011-10-29, 13:29   Link #4160
ars89
One-Eyed Dragon
 
 
Join Date: Sep 2010
Location: NJ, USA
Well not even Kuroneko could take down Kyouko. At least a Madoka girl will lose this time. Rooting for Madoka.
__________________
ars89 is offline  
Closed Thread

Tags
2011, saimoe, tournament


Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -5. The time now is 22:07.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.
We use Silk.