Forum - View topicAnime Expo Cosplay Contest
|
Goto page Previous 1, 2, 3 Note: this is the discussion thread for this article |
| Author | Message | |||
|---|---|---|---|---|
Sunday Silence
 Posts: 2047 |
|
|||
|
Is there something wrong with the voting system? I get the black guy from FFXIII every two votes I do for some odd reason.
It's like this: Sazh Katzroy vs. XX XX vs. XXX XXX vs. Sazh Katzroy XXXX vs. XXX Sazh Katzroy vs. X XX vs. XXXX Sazh Katzroy vs. XXXX Last edited by Sunday Silence on Mon Jul 11, 2011 12:34 am; edited 1 time in total |
||||
|
||||
The American Average
 Posts: 644 Location: Jehuty |
|
|||
|
they're all pretty good I'd hate to be the judge.
|
||||
|
||||
tuxedocat
 Posts: 2183 |
|
|||
Yes, this! I thought those Berserk cosplayers were really good. It looked to me like they spent a lot of time on their costume. But I noticed that when I was voting on day two, I kept getting the same cosplay girls in corsets fed to me in the "random" votes. I was only given the choice of the Berserk cosplayers maybe 3 or 4 times, and I up-voted them each time. The voting seemed so wonky in the way I kept having to vote on certain people over and over again, I decided not to vote anymore. |
||||
|
||||
bglassbrook
 Posts: 1243 Location: Gaithersburg, MD |
|
|||
Because apparently there is no easier system of ranking them then by voting on every-single-possible-combination head-to-head. Granted is forces someone to actually consider them against the others, which could cause better results than a flat out single-shot positioning... it just takes FOREVER (and since you don't know in advance that is the "normal" method, it makes you wonder if the site is broken.) Also sounds like there were some interesting options that didn't make the finals. |
||||
|
||||
Tempest
 I Run this place.ANN Publisher  Posts: 10420 Location: Do not message me for support. |
|
|||
|
We have an alternate algorhythm for the votes that pairs them off in a more efficient manner to reach a top 5 quicker. I'll use that method next time to see if people prefer it.
-t |
||||
|
||||
R315r4z0r
 Posts: 717 |
|
|||
|
I think this voting system is crap. (No offense meant to the person who made/implemented it.)
I thought the best cosplayer, by far, was the one girl who went as Link from Zelda... but after voting through each person she wasn't even in the top 5 of my votes?! That doesn't make any sense... I voted for her all 3 times she appeared in the choices and yet the system calculated 5 other cosplayers as my top 5? And it isn't even just that... but the ones it chose as my "top 5" aren't even close to the ones that I would have picked... In fact, for 4 of the 5 cosplayers shown in my top five I always voted against... so how is it they got voted up?! Why couldn't they just give us a flat out album and then just have us pick out the 5 we liked the best? Wouldn't that have been easier and more to the point? Just because you might like A more than B and B more than C doesn't necessarily mean you'll like A more than C as well... It all depends on what the voting options are. I might like the detail of costume A better than the detail of costume B. I also might like the effort put into costume B more than the effort put into costume C. But that doesn't mean that the person wearing costume C isn't able to pull off their look better than the person wearing costume A. This voting system isn't broken.. it's just highly unfair. |
||||
|
||||
|
tuxedocat
Posts: 2183 |
|
|||
The same thing happened to me. I was pretty much forced to vote for this one costume over a dozen times, and she ended up as the number one vote getter because of that. The algorithm kept feeding her image against others over and over, so she took first, even though my favorite was clearly the Berserk cosplayers (who came in 5th!! ? !), whom I only got fed by the algorithm four times.
I don't think the algorithm was designed to be unfair, but you make a good point about how mathematically flawed this thing was, results appearing as if they were skewed in favor of certain characters. A clear favorite 1 though 5 (or 10) voting system would have worked much better. It would be nice to at least try this, and compare the results. I had my husband take a look at it after I spent (way too long) voting on day two's choices. It was just poorly designed. In his words "messed up" (He has designed algorithms for Banks and Universities BTW, so he knows about this.) ---I hope you didn't pay a lot for this algorithm, Tempest. edited to add: My husband suspects the problem might occur because of the "too close to call" voting button which will be clicked mostly for the middle-of-the-pack choices which are neither best nor worst. Since the algorithm can't recognize those as favored choices, it feeds those middle numbers repeatedly, searching for a definitive result. By this reasoning, the favorites have been chosen from the median. urgh. 
Last edited by tuxedocat on Mon Jul 11, 2011 11:34 am; edited 1 time in total |
||||
|
||||
|
Sunday Silence
Posts: 2047 |
|
|||
|
I give up. I keep getting the same photo of some of these cosplays multiple times in a row (I got Panty and Stocking as one of the choices 7 TIMES IN A ROW and the Suckerpunch girls 9 TIMES IN A ROW!!), and it looks like that the system is cycling thru the same 7 pictures, with maybe the others thrown in for an occasional chuckle.
|
||||
|
||||
|
Lightning Leo
Posts: 311 Location: Earth |
|
|||
|
Hmm, judging from the comments it might be a good idea to revamp the voting system, seems there are some underlying flaws in its mechanics (I imagine it would really be a pain, though). Easier said than done, right?
 But, with respect to the sincere efforts of these cosplayers, and to more accurately and legitimately reflect the voters favorites, it seems reasonable to hold a recall with a revised system in light of what appears to be unbalanced vote allocations.
|
||||
|
||||
|
tuxedocat
Posts: 2183 |
|
|||
|
I kept getting a cosplayer whom I will call Girl B fed over and over. Girl B's cosplay was fairly good. I would have ranked her towards the top of the middle-pack. However, I was never given the choice to vote for Girl B over my top 3 or 4 choices. She ended up as my top choice, which was totally wrong.
The algorithm obviously didn't understand the difference between "too close to call" (unfavored) and "too close to call" (favored). Every time the "too close to call" button is clicked, the algorithm is, I suspect, assigning the two choices a positive value, despite being favored or unfavored by the voter. ie.: I am given the choice between to middle-pack choices and I click "too close to call". The algorithm assigns both photos a value of +1. A clear up vote is going to assign a value of +2. However, +1 is not a definitive choice, so those photos keep popping up as choices. The accumulative effect of positive values bumps middle choices up into the top tier. The more you click "too close to call" the more the photos in those choices are awarded a positive value. (our theory) I remember the NBAA voting system was just a clear list, where the voters were ranking their choices numerically. May seem very rudimentary, but it works. This voting system used for this coplay contest not only produces wierd results, it was also extremely cumbersome and time consuming, and the results suspect. |
||||
|
||||
Mad_Scientist
 SubscriberModerator  Posts: 3011 |
|
|||
I think the top 5 shown at the end are the current top 5 as calculated by everyone, not your top 5. Could be wrong. But I agree that the voting system is a bit odd. |
||||
|
||||
|
Sunday Silence
Posts: 2047 |
|
|||
Just a simple "Here are the contenders, click the button next to the pic to chose your favorite" would've sufficed. |
||||
|
||||
|
bglassbrook
Posts: 1243 Location: Gaithersburg, MD |
|
|||
That was the conclusion I came to (considering how many were at the bottom of my list.) Your personal top-5 updates itself as you vote... the left-most images. Which is helpful in that you can make sure not to throw-away vote the same one too many times. For anyone who reads the boards first, how does it handle the "too close to call" votes? Half each or no points? (You can hover over the picture to see the number of votes, or current contenders' ranks at the end.) |
||||
|
||||
|
SaiyaKaizoku
Posts: 2 |
|
|||
|
I agree with the consensus that this voting system should be redone. There were too many times where I had to throw my vote away just to be able to get a chance to vote on my favorites. A simple numbered ranking system would have been better.
Although, I will say that I'm glad that Hyunkel ended up as a finalist. One little nit pick, he should have chosen to either hold his sword or wear it on his helmet. |
||||
|
||||
 |
All times are GMT - 5 Hours |
|
|
|
Powered by phpBB © 2001, 2005 phpBB Group