I think that the voting wording should be revisited, the word good a a ten would imply that half of the photos would get a ten, I think 10 should be execellent or superb or something and that 6 should begin above average. Most of the photo's are good, many are very good, some are excellent and eevery once and a while there is one that is a grade above all the others that the scale does not accomodate that opinion. Just my opinion, thank you.
sounds fair enough to me - as 10 divides equally into five - how about 0 being Bad, 2 Average, 4 Moderate, 6 Good, 8 Very Good and 10 Excellent - makes things a little easier to quantify
90% of the 10,000 images already have disproportional ratings and changing the voting now would just screw the system even more.
There is also the fact that voting would still be as biased as ever. Changing the word next to the number wouldn't change that.
The best rating system I've ever seen is at gfxartist.com. they don't get a % rating, they get a gross rating (i.e. total of the sum of all votes). I believe (if I can remember it right) that regular members get to vote between 0 and 4, paid members get to vote between 0 and 6, and mods get to vote between 0 and 8. That way, biased members would have to be mods or pay to get a high voting privelege. If you want proof that it works, go there and see for yourself.
I believe the only reason the word "Good" is there is to show that a 10 is the highest rating, and 0 is the lowest, so people know how the scale works. Personally I think the voting system should stay as it is, there are too many images on the site that have already been voted on using the old system, and besides the system works as well as you could reasonably expect. If you beleive there is a bias in how people vote, that bias isn't going to change just because a different voting system is put in place. Remember, the c-index is just a number, everyone has different opinions and just because you like or dislike an image it doesn't mean everyone else will feel the same. Also remember when the mods move images into the permenant galleries, they use their own judgement; the c-index is just a guide.
it still doesn't justify the fact that the images that are getting the most exposure aren't the best images.
If it is in fact true that the c-index doesn't matter, why not make the rating system based on views and downloads only? No matter how biased people are, they will still only look at images that they like, therefore reversing the current system. As it stands, the highest rated images receive the most popularity. If u based it on # of views and downloads, the most popular images would receive the highest rating (i.e. reversing the system).
This change would be easy to implement since all the images already have a view and download count.
Not necessarily... if I like an image, I may check back on it often to see the comments, current rating etc. several time, but I will only d/l it once... therefore the # of views will be significantly higher than the d/l's...
Actually, if you sort the images by d/l or by views, almost all the images at the top of the list are rated 99 anyway, so it probably wouldn't make much difference. Another problem with using that system is that the oldest images are always going to be at the top of the list (as they are at the moment). And as for the bias thing, if people are bias toward a particular artist, they are going to check out any new uploads from that person first, and then go through the others if they have time. So the effect would be the same. If this bias does exist, there probably isn't any way to eliminate it; just remember, it isn't a poularity contest. If you post a good image, you can take satisfaction in knowing yourself that you did a good job. Even the most popular artists on caedes have posted images that don't rank well, the important thing is not to let it discourage you...
I know the images w/ most views or downloads have the 99 c-index - because they get viewed more due to their high c-index. And I know that c-index takes views and downloads into account.
So since voting seems to be the only factor of the c-index that doesn't work due to bias, why not take it out completely and leave c-index up to the other factors? I think we'd start seeing better results in the ratings if we'd get rid of voting (or at least seriously rethink how it works).
And Sam - from that last post, I'd say you don't believe the voting around here is biased. Yet some of your experimental bryce works and terragens are part of the millions of bryce and terragen renders on the net that look EXACTLY the same (i.e. green mountains and blue water, metallic pre-made bryce object in a default bryce sky, etc.). And these are the images that are getting 99 ratings while n00b work that kicks the pants off of any of our work is sitting at the bottom w/ 50's and 60's.
The biggist reason for all the bias isn't the authors name tho - it's inexperience on the part of the viewer. On one such image, sum1 called you "The King of Bryce" (and it was a well-known member too). I laughed to myself when I read it because the image probably took 3 minutes to model. If sum1 were to actually look at some real bryce and terragen galleries on the web, our images would be getting 1's and 2's - not 9's and 10's
I would be surprised if bias doesn’t exist. However the sheer number of votes given to most images will rapidly cancel the bias. I believe it takes a couple of weeks for the c-index to reflect the true value of any image.
There are 9991 images on the site by today’s count, changing the voting method would create a massive problem for the existing images.
I don't believe anyone ever called me that Sam... Maybe you are right and there is a bias, but there's no point fretting about it. I think voting is the only way to rank images, as I said if you ordered them by number of veiws or wotes, the images that have been on the site the longest are always going to be at the top of the list...