Go Back   Electric Mole Forums > Shiina Ringo & Tokyo Jihen > Main Forum
Register FAQ Members List Calendar Search Today's Posts Mark Forums Read

Main Forum The place for general discussion. Old news and speculation, polls, trivia, memorabilia, favorite songs, and so on.
Reply
 
Thread Tools Search this Thread
Old 2009.02.06, 09:21 AM   #201
digdad
Senior Member
 
digdad's Avatar
 
Join Date: Dec 2007
Location: Tampa, Florida
Posts: 265
digdad has had more than 15 seconds of fame
Default

Originally Posted by D.K.Liu View Post
Based on what I've read so far of this thread, people have mostly based their ratings on anecdotal evidence: 1.) saw many people praise song A, I don't think it deserve it -> overrated, 2.) don't see many people praise song B, I think it's great -> underrated. After going along this line for while, people gather enough data to say things like: many people think song A is underrated, but it's quite mediocre -> overrated underrated song etc. which is somewhat amusing to read.

Intuitively, a situation where the majority thinks a song is underrated/overrated is nonsensical: since most people agree with each other they should believe that the song is given its proper rating, only the minority has the right to stake the claim that a certain song is overrated/underrated. When we label a song over/underrated, we are basically disagreeing with the majority opinion. The problem we have is that we don't really 'know' with great certainty what most of the fans think.

Another thing to think about is scope: whose opinion do you want to comment about? You may not care about what the general public thinks about Superstar, but do care about what other EMF members think.

Polls are one way to gather information about people's opinion, however, with such a large number of songs, one person one vote would not provide sufficient information about people's preferences. I suggest that we gather information about people's opinion using the Condorcet/Schulze method.

It's a preferential voting system, so we'll have to rank the entire Ringo catalogue (the system allows for 'ties' and 'no opinion'). I know it's sort of a large undertaking, so I don't know how enthusiastic people are about this.

If you are worried how to implement the vote, a guy at Cornell University has setup a free Internet Condorcet Voting Service that makes it easier to conduct such a vote. There are two modes: public and private elections. The public one is one IP address per vote, while the private one requires voters to provide their email address. Any suggestions as to which one we should use?

Maybe I'll pique your interest by starting with a simpler task of ranking SR/TJ's albums first?
Edit 1: I know this is off topic, but you can go here to vote on your preferences about SR/TJ albums.
I'm trying to think of how a ranking of songs would work. If I understand this system correctly, each song is compared to every other song, meaning we would have to treat the entire SR/TJ catalog as a set to be ranked. That is quite a daunting task, and not one I think I could do. In my mind it would be preferable to rank more manageable subsets (such as songs on a particular album) then combine the results somehow. I don't believe this method would allow that, but I'm not certain (while I have a background in statistics, I am not up on voting systems and methodology).

I've participated in a similar poll, albeit manually tallied, on a Beatles forum where we undertook a sort of playoff system. The end goal was to wind up with a number one rather than a complete rank though. What we did was rank the songs on a single album at a time. When all the albums were internally ranked, the three top songs from each album moved on to a second round. They were split into groups of twelve or so (not sure how, but this is certainly where you run the risk of bias creeping in) and those groups internally ranked. We eventually got down to a single grouping of 10 top songs and a final ranking was created. This method took a long time to complete but was very manageable, not to mention a fun diversion over a long period of time.

Certainly drawbacks exist if we are looking to create a robust ranking, but it is a fine method to determine an overall winner (and could be used to determine an overall loser also if extended to the bottom tiers). A logical ranking drawback involves the fact that subsets are equally ranked which may not be the case. In other words, all albums are considered equal from the start, while one album and the songs on it may be considered decidedly weaker than one or more other albums by the voting majority (not naming any names ), but the method ignores this. The top 3 songs on the weak album would go up against the top 3 on the other albums, when most would consider the middle or even bottom tier songs on the strongest album to be better and higher ranked than the best of the weak album. This should not affect the final winner, but clearly alters a final ranking.

Sorry if I just mucked the waters instead of offering a viable alternative to ranking the entire catalog, but I guess my question is do you know of a voting method that would allow the division of catalog into more manageable voting races yet still result in a complete ranking? Although, if we are only interested in identifying truly overrated and underrated songs, the middle may not really matter, just the top and bottom rankings.
__________________
Last.fm | Blog
digdad is offline   Reply With Quote
Old 2009.02.07, 02:13 AM   #202
merman
Senior Member
 
merman's Avatar
 
Join Date: Nov 2007
Location: Bratislava, Slovakia
Posts: 561
merman is an asset to this community
Default

Originally Posted by Jonny View Post
Well yeah but KZK is still overrated... here on the forums... I'm not sure how people on the whole (world) rate it. I guess in Japan, the general Joe, most people just remember songs from MM/SS.
That sounds like what's more popular is objectively better, and that's crap, because then Ayumi would have better music than Ringo.
For a small sample of how people worldwide rate it, take a look here:
http://rateyourmusic.com/release/alb...kuri_no_hana_/
merman is offline   Reply With Quote
Old 2009.02.07, 06:18 AM   #203
D.K.Liu
Senior Member
 
D.K.Liu's Avatar
 
Join Date: Jun 2008
Location: Melbourne, Australia
Posts: 143
D.K.Liu is an asset to this community
Default

To Ringo~Bingo:
I wasn't sure whether to include Watashi to Houden either but since the system allows for ties, I thought it wouldn't hurt to include it.

To digdad:

I haven't looked deeply into voting methods either, but just off the top of my head maybe Borda count or some sort of point based system would allow us to divide the votes into smaller elections?

Actually on second thought a point system could be used to replace the entire vote... Each person rates every SR/TJ song on a scale of 0 to 10, we then use the average score to rank the songs.

Earlier today, I was thinking along similar lines, but in the context of making it practical to use the Schulze/Condorcet vote for the entire SR/TJ discography. I was thinking about devising a point system to help each voter rank the entire catalogue, which can then be used as inputs to the Condorcet vote. Now I wonder if we would get a different ranking if we just sorted the songs by their average scores vs using each voters score as inputs to the Condorcet vote...

Maybe I'm just making things complicated, but I was considering for the point system to get the voters to judge each songs using multiple criteria such as: innovativeness, instrumentation, vocal, lyrics (would be hard for us) etc. I am not very experienced in thinking critcally about music... anyone has any suggestions in this regard?

I agree that if we just wanted to provide data for the purpose of this thread, doing something like a top 5/bottom 5 vote would save a lot of work over the above methods.

P.S. so far 5 people have voted... perhaps there is a way to notify EMF members of the existence of a poll through email...

Last edited by D.K.Liu : 2009.02.07 at 06:32 AM. Reason: update on the poll
D.K.Liu is offline   Reply With Quote
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

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

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT -8. The time now is 11:09 PM.