Fanboys purposely giving low rating to specific releases?

Forum for discussing AniDB rules & standards. No small talk!

Moderator: AniDB

SinsI
Posts: 20
Joined: Sat Apr 15, 2006 8:25 am

Post by SinsI »

Since almost nothing deserves 1 or 10, it is a very safe bet to cut them altogether. If it is really good or really bad, and not just boosted by fans/haters, even without those votes it would be placed high/low enough. Just take a look at the aforementioned Ergo Proxy groups:
Ani-Kraze: 7.96, 53 votes (With 1 and 10: 7.94, 118 votes)
Shinsen: 7.74, 68 votes(With 1 and 10: 8.02, 155 votes)
Both are statistically reliable (more than ten votes), both are at least good, so to choose you have to read comments and select the one that suits you better.

The only problem is the manual re-calculation. It would be veeery nice if such a feature was added to AniDB(e.g. in profile - set minimum included vote to X and set maximum included vote to Y).
exp
Site Admin
Posts: 2438
Joined: Tue Oct 01, 2002 9:42 pm
Location: Nowhere

Post by exp »

SinsI wrote:The only problem is the manual re-calculation. It would be veeery nice if such a feature was added to AniDB(e.g. in profile - set minimum included vote to X and set maximum included vote to Y).
impossible, always keep in mind that whatever we decide to do needs to allow good caching.

BYe!
EXP
Klutz
Posts: 42
Joined: Sun Jun 06, 2004 5:59 am

Post by Klutz »

/bump! hmm not really first time posting in this thread. I can't believe how some people are thinking. First I agree that those 1/10 group votes are stupid, but there isn't really anything to do about it. There is in fact the possibility that you actually LOVE or HATE whatever release a specific group has made. If someone votes 1 if he hates a person in the group or had a bad experience with one of their earlier releases, well then at least his single vote will not have too much impact on the overall.

Some people then suggested to remove the 1 and 10 votes alltogether. One big LOL from me. There's always gonna be two extremes in a vote system, if you remove 1 and 10 then people will vote 2 and 9 instead and so forth. In the end we will have one pissoble vote left. How you like this release 5? or 5? or possibly even 5?!?

In the end however I think the up/downvoting on animes instead of specific groups is a much bigger problem. If you see a popular release (especially those animes in top10 and those right below) then this anime can be as brilliant as it wants, it will surely have a solid bunch of 1-votes. It's so very sad. How can ANYONE vote 1 for MONSTER HOW???
RealFduch
Posts: 4
Joined: Mon May 08, 2006 7:57 pm

Post by RealFduch »

BTW, I saw someone told that 1 and 10 kill each other. That's obviously not true for good anime. 1+10 skew rating towards 5.5

I think that rather simple function can be thought of to determine extreme votes and voters. Voters can can be identified by the cumulative statistical interpolation error the bring.

I don't think that I want seeng people with high percentage 1/10 votes. I dont have monochrome mind. I'm not living in imaginary magical world where everything is black.

I don't belive in numbers such as:1 - 15
2 - 0
3 - 0
4 - 0
5 - 0
6 - 7
7 - 14
8 - 23
9 - 30
10 -19

If there are 15 extreme people that think that it's worst anime/release, then there should be even more voting 2.

I think that the should have right to vote what they want. But want to have the option not to see them (voters).
Grv
Posts: 2
Joined: Sat Aug 19, 2006 12:26 pm

Suggestion on limiting the impact of "fanboy-ism"

Post by Grv »

Right, I've asked about whether or not an official thread exists for this topic and as I understand it there is none.

This subject is incredibly annoying to most people I imagine, especially to anidb mods by now, given the amount of feedback they get upon this single topic.

My suggestion for "rectifying" the situation would require some modifications to the website code.

In essence:

If a user chooses to vote a particular fansub group or show that a group has released between 1 and 4 or they have voted 10, they should be prompted to leave a comment justifying that action, if there is no comment left, the vote is nullified.

Now I appreciate this is extra work and potentially extra strain on the hosting, I also appreciate it is not going to be a 100% success rate.

However, if one considers the skew between a 5 down-vote and a 9 up-vote in comparison to the skew between 1 and 10, we should begin to see significantly better and accurate scores.

It is disheartening for small groups when they are down-voted, which is a shame as this is a for-the-fans-by-the-fans community, and everyone who wants to sub should be encouraged and be given constructive criticism.
Hoerie
Posts: 79
Joined: Sun Aug 07, 2005 12:15 pm
Location: Delft

Post by Hoerie »

And then you'll supply anidb with the tard-proof code that'll evaluate those comments for validity? That will only move the problem along a bit...
egg
Posts: 769
Joined: Tue Nov 11, 2003 7:17 am

Post by egg »

Hoerie wrote:tard-proof
My father has a saying, "You can't make anything idiot proof, idiots are too inventive." Same thing applies to tards.
Hoerie
Posts: 79
Joined: Sun Aug 07, 2005 12:15 pm
Location: Delft

Post by Hoerie »

You should listen to your dad, he says sensible stuff... ;-)
Neko-Musume
Posts: 56
Joined: Wed Feb 02, 2005 2:31 pm

Post by Neko-Musume »

Heh, the way I saw it (on a bumper sticker) was "Make it idiot-proof, and someone'll make a better idiot." But I think I like your dad's better. :D
Grv
Posts: 2
Joined: Sat Aug 19, 2006 12:26 pm

Post by Grv »

Hoerie wrote:And then you'll supply anidb with the tard-proof code that'll evaluate those comments for validity? That will only move the problem along a bit...
I think without a shadow of a doubt, more people look at the scores and then leave the site than those who actually go in and read the comments. The problem is a very hard one to resolve due to the culture these days, in terms of evaluating comments you could filter them via a number of rules, and filter out yet more "tards", yes.

As I stated, no "solution" will ever be 100%, but as it stands the problem only appears to be escalating.
AnimeOtaku
Posts: 158
Joined: Sun Jan 09, 2005 10:54 am
Location: Germany

Post by AnimeOtaku »

An idea:

In sports there are votes too.. from I think 10 people who vote, the lowest and highest rate will be removed from the end score.

My idea would be at a minimum of 100 votes for anime and maybe 10 for releases, to "disable" the highest and lowest 10% of votes for the end score
Hoerie
Posts: 79
Joined: Sun Aug 07, 2005 12:15 pm
Location: Delft

Post by Hoerie »

It really surprises me how much people want the ratings to be 'correct', to me the ratings don't mean very much. Other than animes scoring <6 being possibly mediocre or worse, I don't really care if an anime scores a 6.3, 7, or 8. If you want you can look at the bar-graph and draw your own conclusions. The current system with all votes being equal is simple and understandable for anyone. Imposing all sorts of statistical stuff on the data just gives people a false sense of 'correctness' imho.
AnimeOtaku
Posts: 158
Joined: Sun Jan 09, 2005 10:54 am
Location: Germany

Post by AnimeOtaku »

I'm just want to find a way which benefits all sites the best.

Of course, you can't do it right for all people.

Btw it's not a statistic, it's more like a poll.
zaroba
Posts: 1
Joined: Wed Sep 13, 2006 5:23 pm
Location: Hereford, PA
Contact:

Post by zaroba »

thiers a (probably) much better way to go about this then eliminateing votes.

what if thier was a limit on how many times a person could vote on an anime episode? or if they could only rate one group? then the terds woulden't be able to vote thier group a 10, and all the others a 1. nor would they be able to go into an anime episode and rate the file that was subbed by thier group as a 10 and all the other files as a 1.

it'd be like saying 'vote only for the group you have'


side effect of this would be that every group would probably only have positive ratings. to get around that issue, base it on the % of total votes for all the groups. then if one group 1 has 75 votes, and the group 2 has 25 votes, group 1 would have a 75% preference rating and group 2 would be 25%. could do the same for the episode ratings.
Locked