All right, here's another idea. It seems like the problem with our rating system is that some people are rating too high and some people are rating 0. Mostly the latter.
This has a significant effect because TDC currently calculates it by mean, if I'm not mistaken. Examples of votes if they were calculated by median instead:
Good game: (0,0,0,2,4,4,5,5,5,5)
Bad game: (0,0,0,0,1,2,2,3,3,5)
Great game, with idiots voting against: (0,0,0,3,5,5,5,5,5,5)
Good game: Mean = 3 stars, Median = 4 stars
Bad game: Mean = 1.6 stars, Median = 2 stars
Great game: Mean = 3.3, Median = 5 stars
Sooo.. long story short, a good game will have a good rating. A bad game will still have a bad rating. A great game may have a bloated rating, but could be balanced once more people start voting.
Under the present rating system, the good game will look bad, and the great game will still look bad.
The current downloads rating system does work well because nobody's going to look like an idiot by giving the game a 0 star rating, but it seems to be messing up for articles, because you can't force them to comment to rate, but it lets them rate without shame.
Edited by an Administrator.
Disclaimer: Any sarcasm in my posts will not be mentioned as that would ruin the purpose. It is assumed that the reader is intelligent enough to tell the difference between what is sarcasm and what is not.
Mmm, can be a little unfair at times. I like the 5 star rating system. However it feels a little limited as to the rating you can give. Kinda like you can't fully express your overall feelings towards the game. 10 stars would take up too much room but, if it went up in halves then it would give us more stars to play with.
I don't really think we need more. Reviews let you give more detail if that's what you want. Lol, nobody uses reviews anymore.
Disclaimer: Any sarcasm in my posts will not be mentioned as that would ruin the purpose. It is assumed that the reader is intelligent enough to tell the difference between what is sarcasm and what is not.
The problem with these user ratings is that, there is no specific value to the stars. If you are trying out a game, and it is quite a good game, but not a super game. You want to give it 3 stars, but to avoid misunderstanding, you give it 5. The point is; the game is good, but not super. So if there could be giving more specific values to the stars, f.ex. 5 stars = EPIC (means good), 4 stars = SUPER(means good), 3 stars = GOOD(still good), 2 stars = AVARAGE (the game is ok) and 1 star = NEEDS ALOT OF IMPROVEMENTS. (bad or not quite finished)
Lol, I don't think that's the real problem. People pretty much know that a really good game is 5, an OK one is 3, and anything less than that 0-2 or not deserving of a vote. The problem is when some people start voting 0 to be mean and others start voting 5 to cover up for the 0s.
Disclaimer: Any sarcasm in my posts will not be mentioned as that would ruin the purpose. It is assumed that the reader is intelligent enough to tell the difference between what is sarcasm and what is not.
how about only letting people see the game's overall rating after you've voted. That would go some way to stopping people voting 0 just to bring the average down or 5 just to raise it.
Of course that only works if they're people that see the score and think the average is higher or lower than what they think it should be.
That would kind of mess up the whole idea behind the ratings. If you can't see the rating (or sort for that matter) for a game before downloading it and rating it, many people won't bother downloading many of the uploaded games I believe. It would be hard for visitors to find those old gems if they weren't allowed to see the ratings.
Ok, sorry for my messed up grammar and such, but it's 5:51 am over here and I'm tired.
I don't like the way you presented your examples, Muz. I'll just simplify them (so I can do my maths better).
Good game: (0,0,0,2,4,4,5,5,5,5)
Bad game: (0,0,0,0,1,2,2,3,3,5)
Good game, w/ odd voting: (0,0,0,3,5,5,5,5,5,5)
What if the rating system used variance? Using the program I wrote a few years ago for my calculator, I think it'd be a more accurate system. Something to note is that my program outputs two numbers, and I'm not really sure what they mean, so I'll just report both of them. If anyone with a better sense of math can run the numbers and get the same results, then please tell me what the parenthesed numbers are. Anyways:
The good game would have a rating of 4.61 (or ~2.1471). The bad game would have a rating of ~2.8547 (or ~1.6896). The "oddly voted" game would have a rating of ~5.1790 (or ~2.2757).
Consistent voting patterns are definitely better for variance, it seems.
Median is a bad and often misleading statistic. Mean is fine I think for now, almost every other website in the world with ratings uses mean.
I don't think you can really use outliers - as well as being complicated to do, outliers only really apply to large amounts of data with one or two values which are very far away from the rest. One rating of 1 for a game with an average around 8 wouldn't alter it that much, it's only when large amounts of people vote 0 for a game, in which case those votes cease to be outliers.
I think making halves available would be better, to differentiate between goodness and greatness.
You should not be able to rate something 0, because that is the same as not rating the game at all. The lowest possible score that one should be able to give is a 1. That way when someone gives it a bad rating, perhaps on purpose, the overall rating will not drop as bad.
(0,0,0,2,4,4,5,5,5,5) Average= 3 star rating
(1,1,1,2,4,4,5,5,5,5) Average= 3.3 star rating
Now the difference may not seem like that much here, but on a larger scale it would make a big difference.
The main problem is that we will always have idiots who rate too high or too low just for the heck of it. What we really need here is a "is not an idiot" status that allows you to vote and stuff.
I don't know if it's all that important to make sure people vote properly. If your game gets one 0 rating from someone you can ignore it. If you have 100% 0 ratings then maybe you should start to take notice.
I mean how many people play a game extensively before making their rating? Maybe the problem is people care too much about what score their game gets?
Advertisement
Post Reply
Edited by an Administrator.
