Dmitriy 2013年3月5日下午4:46
How are rating stars calculated?
I have just checked out the new "Item Stats" page and now I see how many people voted up and down for my item. I get:
Total ratings 30
Positive ratings 28 (93% of total)
Negative ratings 2 (7% of total)

Workshop shows 3 of 5 stars rating. Why? I thought that those stars are directly related to users' positive\negative ratings.

http://steamcommunity.com/sharedfiles/filedetails/?id=125888855&searchtext=

Wowza 2013年3月5日下午4:52
I think the number of ratings are also a factor.
TomB  [开发者] 2013年3月5日下午9:33
Wowza is correct in that the total number of ratings is factored in. We use what's called a bayesian average (see http://en.wikipedia.org/wiki/Bayesian_average).
Dmitriy 2013年3月7日下午2:05
This is great for making a list of most popular and highest rated items. When user clicks "Top Rated All Time" he gets the list of really good items and not some new items with 10 votes and 5 stars.

But when it comes to looking at the individual items it is a bit confusing. I'm saying about my own experience. When I see a map that has 30 votes and 3 of 5 stars I make a conclusion that this map is something very basic or not yet finished. I don't know that those stars are bayesian average, I think that it is just an average rating, a relation between likes and dislikes. So I don't waste my time trying it out, especially if there are higher rated maps.

I think it would be great to implicitly use bayesian average for the list "Top Rated All Time". And use just an average for displaying stars. Although it may be confusing for people to see a 5 star map being listed under the 4 star map :)
Dmitriy 2013年3月8日上午8:27
Also it can be better to be able to rate a workshop item from 1 to 5. Same like imdb.com rating works
TomB  [开发者] 2013年3月8日上午9:15

Also it can be better to be able to rate a workshop item from 1 to 5. Same like imdb.com rating works

There have been several studies that show 5 star rating systems don't actually give us great data. We want users to make real choice. There is plenty of debate on this topic of course, but we feel like the up/down rating system is the simplest and gives us the best data.

Here are some informational links if you are interested:
jessd 2013年7月21日下午11:35
gotta love the trolls comin in and givin a thumbs down just cuz they don't like the idea or style of the mod. what's with the thumbs down? i can understand if a mod doesn't work, but why allow people to vote a working mod down just because it's not their cup of tea? sounds like a good way to run the modders off. i know it frustrates me....

CS:GO Cataclysm 2014年2月19日下午7:17

Also it can be better to be able to rate a workshop item from 1 to 5. Same like imdb.com rating works

There have been several studies that show 5 star rating systems don't actually give us great data. We want users to make real choice. There is plenty of debate on this topic of course, but we feel like the up/down rating system is the simplest and gives us the best data.

Here are some informational links if you are interested:
I know this is a while a go now, but how many ratings are required for a rating out of 5 stars? :)
Dmitriy 2014年2月19日下午8:34
as guys already said it depends on how many ratings do other item have.
CS:GO Cataclysm 2014年2月20日上午6:11

as guys already said it depends on how many ratings do other item have.
Okay fair enough, wasn't sure if you needed say 10 ratings for it to be eligible for a 5 star rating, with the Bayesian Average being applied after that :P
gtamike_TSGK 2014年7月5日上午1:38

gotta love the trolls comin in and givin a thumbs down just cuz they don't like the idea or style of the mod. what's with the thumbs down? i can understand if a mod doesn't work, but why allow people to vote a working mod down just because it's not their cup of tea? sounds like a good way to run the modders off. i know it frustrates me....

Yeah it's odd you can vote when the person has not even download the item.
Catalin 2014年7月5日上午4:01
smiling Aktheon 2016年12月27日下午7:44
that sounds weird.
- you use 2 star system (tmb up/down) instead of actual 5 star system and than respresent result as 5 star
- you count non voted persons as a people that voted neutrally? That leads to more questions:
a) should we call it a 3star system than? Maybe you add 3rd button to voting system, so people can thumb up/down/neutral ?
b) counting non voted persons meens that people are forced to vote anyway. It's up to game culture and community health to activly vote for mods. Also voting is heavily effected by how often author communicate with subscribers on comment section leading to fact that subcribers are more likely to vote.
How communication represents mod quality? It does'nt.