I'm sick of seeing my highest rated models disappear from their placing because one person decides its only worth 4 stars instead of 5.

When a model has 30 5 star ratings and 1 4 star rating... how is it fair that this model now takes a 4 star rating position when I search by

ratings? A lot of people search by rating not by date... so now I have all these 30+ 5 star rated models lost in the hundreds of pages here..

seriously I can't even find what page some of them have  ended up on...

I really don't feel compelled to upload anymore to this site because of this stupid inaccurate system of calculating  the overall median of ratings.

Insert Image

Explain to me how all of these models with 20+ or more ratings feature after models with 0 ratings?

My building ruins model went from page 1 to page 249 because one person gave it a 4 star rating.

Ridiculous!

I'll show how to calculate the mean star rating of my building ruins model...

29 x 5(stars) = 145

1 x 4(stars) = 4

Add thes together: 145 + 4 = 149

divide 149 by 30(overall ratings) = 4.966666666666667  This is the average star rating of this model.

30 4.966... star ratings is better than a model with just one 5 star rating.

Hi hypnagogia,

Sorry to hear this discussion.  I will investigate this issue and get back to you.

Regards,

Team ShareCG

You are bringing two issues.

The first issue is the display of uploads without rating placed affront of (in the rating sort) the uploads with 5 stars. Those uploads were rated 5, but we were not displaying their rating as long as they not have at least two rating entries. That logic was missed from the original sort and we addressed that issue. Please check it .

Regarding the second issue of the rating itself. Currently we are using simple average.  we are open to any reasonable change and will consider changing the logic if it is doable and fare to all members.

I see you have changed something and it seems to be better than what it was.

I suggest using an accumulative points system... where 1 star rating = 1 point.

Example: something with 5 ratings has a point value of the sum total of those rating points... 5 + 5 + 5 + 4 + 4 = 23 points. Instead of calculating the average.

This would also fix to a degree things that were rated low either by accident or maliciously.

Thanks

Thank you,

However there is an issue with this logic. If we use that logic may bring a low rated upload with many resposes affront of a high rated high rated upload with two resposes.

For example, one upload has 15 with 1 star rating and another upload has two reposes with 5 starts. Since 15x1(=15)>2x5(=10), the first upload will be rated higher.

So this logic will not work.

Nitai

Yes thats true, but nothing on this site has been rated with 15 x 1 stars.. most people don't rate models at all if they are worth just 1 star.

At the moment the system is flawed and very unfair to all members who contribute to this site.

To me it doesn't make sense that something that has  100 5 star ratings and only 1 4 star rating will always be placed after something with only two 5 star ratings... does that sound fair to you?

why not place them together in groups of how many ratings they have.. and then order them by average.

Eg, all moedels with 20 ratings are placed together but are listed in order of the average.

Do you understand what I'm getting at?

You are giving power to just one person to completely ruin an artists work in terms of its status. I've had someone think he was rating me 5 stars by selecting the 1 star option... he had no way of correcting this.

Anyway thought I would bring this up cause its been pretty annoying to see this happen. I guess it doesn't bother anyone else except me.

Makes me feel like removing all of my work and never submitting anything here again.

Instead of a simple average why not implement a median rating? That would discount very low and very high number and come closer to what people true think it is rated.

Potential problem if there are a lot of votes it may take time to sort those in order to take the median. Perhaps it might be faster to find and discart the highest and lowest votes then do simple averge for what remain. If less then 3 vote then just do simple averge.

Josea,

Thank you for your suggestions. As always we trying to work with our customers to better ShareCG.com.

Rating by mean average is a good idea, but as you indicated just calculating it on the fly for thousands of upload (and their many rating) will take long time and delay the display of the results. To make the change correctly, we will have to develop a system that assign the mean average to each upload when the rating happened.

I'll put it on my to-do list and hope to get into it soon.

Thank you again

Nitai

I agree with hypnagogia, the classification system is unfair and discourages authors we contribute to this site. Hypnagogia is not the only one that this bothers.

Lately, a new member, has been commenting and thanking my models, and then classifying them systematically with 4 stars.

I suppose you think that makes me a compliment when my models really are relegated to very backward positions.

Most of those who download a model nor comments, nor rate.. For some obscure reason most of those who value do with 5 stars. Then there are the trolls that malice and without having contributed even once, without any explanation classified with 1 star. Or the clueless who are wrong to select.

But there are no models classified by 15 users with 1 star. It looks like an all or nothing system. It does not seem the most appropriate valuation calculating an average that takes into account tenths (or hundredths?).

Given this, I also think that the best classification system is given as first criterion the total number of stars. As a second criterion the lowest number of voters (average better) .As third criterion, better punctuation for the latest model.

Although also I find just the following system: First criterion, tolal number of votes.

Second criterion, average better. Third criterion, best latest model.

Actually, I think this rating system already applies. On page portfolio, if the sort is requested by rating, models are grouped by number of votes and then by average.

For example, in my portfolio a model with 15 valuers and an average of 4.7 occupies the position 19 while in position 20 is a model with 14 reviews and an average of 4.96.

It is a system that you have already functioning for each individual portfolio. It would be to extend it to the general classification.

A greeting:

Pedro

Pedro,

Thank you for your comments.
Your sugguection for sorting the "rating" is a lot faster to the implemant compair with the mean-avegage logic suggested by other members.
We can try it, but before we change the sort-order by rating, we would like to hear from more our members.

Other thought I had is to change it to a "Like" system, so each member has the option to "Like" the upload or not. After all, what is the difference between 4, 4 1/2 or 5-stars.

Lets give our members time to give us some feedback.
Thank you all
Nitai

Nitai:

I've been looking at the rankings and I propose the following experiment: Go to page 169 of 3D models classified by rating (or wherever this model "Organodron City" Heminio). On this page are the a number of models classified with 2 rating, followed by 10 models from 33 to 26 ratings. The last comment in 9 of these 10 models is the "thank you" of the user Mandle.

The same is true on page 170 and 171 with 13 and 11 models of the page. And so on these pages continues to models with 5 ratings.

Mandle always thanking and presumably 4 points to very good models, made by very good modelers. (Herminio, hypnagogia, klausdbg, spacebones, Neil Wilson, Luxxeon, and others)

Get your your conclusions.

My conclusions:

Mandle achieved alone and in less than a month distort the rating system.

Mandle not have us mania to hypnagogia or to me personally. It is a widespread plague.

Mandle is a robot (or a very persistent person) programmed to delay 150 or more pages to any model with more than 5 assessments, and taking advantage of the weakness of an unfair system of classification.

Perhaps someone resentful, that is their models achieve better position.

It may be someone who wants to harm Sharecg, trying to favor mediocre models, and discouraging them all to the point that good modelers go forward to other sites.

I fail to fully understand the reason for all this. But either the user Mandle think you have a problem with ShareCG, or ShareCG have a problem with Mandle.

In view of the activity of this forum, I would not expect many comments from members. Perhaps the first comment is a massive defection of modelers to other sites.

Thanks for your patience.

Pedro

Mandle has ruined nearly all of my models now... he did this over a period of 3 weeks.. and is still going. I've tried to contact Mandle to ask why he is rating all of my models with 4 star ratings... no reply!

Like Pedro has mentioned he is doing this to all the top contributing members.

Eventually when you sort by ratings all you will find are models with 2 stars.. I predict all the top models will be gone in 4 months if he continues like this.

I will NOT be sharing anything new on sharecg again until this is fixed. This upsets me because I worked hard on these models and made sure they were of the highest quality... and I love to share them with people who really appreciate them... so its a shame that the rest of sharecg's members will now have one less modeler to share with them.

This person is either being purposefully vendictive or Is a Bot designed to lessen the first sight appeal of sharecg's website.

All,
I just changed the browse page "by rating" sort order to be by "median" instead of average. Following this value, it will sort it by the number of rater. Now "Organodron City is sorted to the first page.

Be aware that we could not afford to make such changes on all the time. No one support us financially nor willing to contribute.

If you to help our effort by spreading the work, it would be appreciated.

Nitai Fraenkel

Edited by TeamSharecg on 09/27/16 04:01 PM.

Thank you Nitai this is fantasitc. Median ratings are definitely fairer and a more accurate representation of the work. Now we can see all of the great work that had previously been hidden/lost in the hundreds of pages of work.

Thank you for listening and implementing this change. I look forward to uploading more work in the future. 

Nitai:

Definitely the classification system applies, seems much more equitable.

Thank you very much for your attention, understanding and for the work done.

I look forward to sharing more of my models, to the extent of my ability.

Pedro

You are all welcome.

Nitai