Introducing new community rating system

thank you for this answer Ed.

Personally, I think that only the purchaser should be able to rated a product, it is the only one to have a correct view ... But that's just my opinion.
Expect to get a lot of ticket seller angry because of unjustified ratification :)
be vigilant :)

the rating system is a real poison. I 'v see these perverse effects on TurboSquid. Vendors rate together, to obtain a maximum rating, regardless the quality of the product., or put a bad note when a product is in competition with them, especially if the product is cheaper.
I speak from experience, it is a very bad idea.

I will just copy this comment of mine from the CGT blog. I want to see what others think before I add anything else.

"Excelent. The rating system will help allot. There are a few things you can do to make it work even better.

First, it seems that cg models and 3d print models are mixed together..?

Then. There is one rating for it all - topology, presentation, render...
Asset got a 5. Was the presentation not good enough, was there only one image, was it not modelled well, what was wrong with it? Ratings like this only separate bad from good with no feedback. Ok, it will filter through the library real quick, what's after?
Rating on separate aspects of a model would give the seller a chance to improve on parts that were not good enough, it would show everyone what is a good asset, it would create a standard. Now it is just a game where you are encouraged to guess the most popular rating rather than to give your own opinion.
For example I know that a product with just one presentation image is no use, but someone else might like the one image and give it a 10...

Also, the description, tags (relevant, irrelevant), even price could be rated upon."

Sounds great to me, will test it extensively. Thanks for the new tool.

captainmarlowe, Audrius,

Thanks a lot for your feedback. This is very useful and we are already preparing specs for phase 2 of this feature.

What we understand clearly is that we have to separate CG and 3DP models in rating system. We are considering one of two ways to do it:
- create a tab in rating page so you can select what type of models you want to rate
- ask designers to choose in their dashboard what they are into: 3DP / CG / Both.
In second case we could use this flag in other places of the site as well.

Regarding incomplete models or the ones you're not sure about, I think we need to add "Skip" button so you don't need to rate model that you're not sure of. Quick tip: you can refresh the page to change the model you're rating.

Regarding "bad ratings to good products" - as Marius mentioned, only approved raters will have impact on final average rating moving forward. Part of this beta phase was to see how many negative cases we have and filter those out so you shouldn't worry about such cheaters.

SuperCigale, thanks for your comment. This is exactly why we are launching it as a demo and will see how it works on our community before letting it do any significant effect. We will monitor raters internally and only the approved ones will be rewarded with reputation and their ratings will make the difference. There will always be a check comparing our own rating with the one that is given by designers so we will clearly notice such over-raters.

So here are my two cents, for what it's worth.
I've tried your rating system today, and let go not even reaching the 50 daily rates.
Why ?
- Because you can't choose between CG and print ready models. I'm mainly a hard surface modeler, and I don't really know about 3D-printing. So I will see an object that seems nice (and may be highly rated by people who bought it and printed it), but that looks to me like rubbish because it is obviously a 3D scan of an object, with an awful topology (on the rare occasions when it's shown) and a simple shader added on it. Which mean that it just took a few minutes to the seller to create it, and is not, in this way, a good model.
- Because you can't skip objects when there is a serious doubt about them. Way too many objects I found today where just with one image, or with no wireframe nor clay render. In this case, my reflex is to rate between 0 and 3 because if the seller doesn't show the wires, it is in my book a big no-no. But, as Audrius stated before, maybe someone rated the model better because he liked the unique render...
I noticed when I tried to rate that each time I had to rate a model with proper renders and wireframes, I was having a good rating score, and often when I rated models for 3D-printing, I was loosing accuracy and finally that I was almost all the time far from the average when models were displayed without clay or wireframe renders...
So as it is, I think the rating system (which is not a good idea in my book, I keep thinking rating should be reserved to buyers or at worse trusted sellers) is not good nor fair.

Again, my two cents.

I have just finished rating the models and I know the rating system is beta version. I gave 9 points to a high quality model but to my surprise its average rating was 0.0 .

Guys, don't worry about the bias or unfair ratings, this is where our internal moderation comes in - we can easily detect such unfair ratings and automatically diminish their weights to the level that it won't matter.

The idea that purchaser is the main who should be able to rated a product on my mind is right.Perhaps it can be taken into account of points in the rate.

I just tried out the rating system. Difficult to judge without including prices, if I rate online purchases this is always a factor especially when comparable items are much lower in price.
I tried to rate based on presentation and fitness for purpose so complex or simple models or type of usage can be judged fairly. For example, a low poly boxy game model can not be rated low because it can't be used in a close up visualisation, or vice versa.
And with end users being everyone from large companies to 3d hobbyists, what may not be suitable for a company presentation may be perfect for a student project, and this is why price would be a factor if I was to rate.

Agree with others, I think the ratings have to be separated into various separate factors. A $5 3D print ring, should never be rated the same as a $200 high poly model with just the same 5 star option.

So, I think there should be some separate factors and one factor being, how much detail is provided in the uploaded model. A model without poly counts, no wires / topology views or low resolution images, should automatically be rated lower for that factor. Perhaps there can be:
Preview rating: automated factors
Purchaser ratings: actual buyers
community ratings: the beta system now

Then average these or weight them into how the marketplace reveals content. Or better, let me search for content based on it. I for one, would only search for high poly models with high preview ratings if I'm expecting to spend >$100. A model at any cost with only one poor image is of no interest to me, regardless of price. Plus I don't care if its been purchased before, if the preview and details tells me enough about the model quality. Not sure how others feel on that, that's my $0.02.

This is the exact reason I have given up playing the game until these issues get sorted. Someone (maybe more than one) doesn't like the game so much that they are sabotaging it by giving bad ratings to good products. I kept finding crazy average ratings, which also suggests to me that these products had only a few ratings. There should be a variable built into code, where if someone's accuracy score falls bellow certain point, their data becomes ignored. Or something similar.
Also I was wondering, is the accuracy score being recalculated once there is more data added? For example there was one rating on the product - a 0, then I gave it a 10, so the average of that product becomes 5, and this "incorrect guess" affects my accuracy score allot, but later on the product gets more 10's as it deserves a 10 and my guess is not so incorrect, but for that to be taken in to account, the score should be recalculated. Well, actually I just realised that this is not that important as "Rankings are based on last 200 votes". Never mind.
Something else that bothers me is that CG and 3D print are still mixed together.
So as I said - waiting for a major updates. ;)

Great to hear from you, Ed.

I think both options (tab and flag) are good, so they can both work together. For example if you choose you are into CG, then you can still vote on 3DP tab but your votes will weigh less in the calculations :)

Thank you for the tip.

I agree with everyone! You need to separate the printed models from 3D models .. Also, I noticed that some very poor models without textures and bad topology have a good rating, why? Some models are repeated several times.

I just discovered this today as it was in the seller's digest newsletter... What's the state of it today, does it work? And is there a way to see the community ratings of your models?