PMOG Forum
civilized discourse for an uncivilized gameworld
(advertisements appear if you're not logged in or need to create an account)
| Author | Thread |
verifex Level 10 
SeerPosts: 17 |
I know you've all been thinking it. If a mission has a broad range of ratings, it usually ends up at 3 stars, which doesn't help anyone decide if it's worthwhile, especially when it seems like too many missions are at that level. I know we've had this discussion before, but I haven't seen any resolution from any side come of it. So, to begin, here's my 2 suggestions:
|
pixielo Level 20 
PathmakerPosts: 3706 |
I believe that missions start out being rated 3 stars, and then go up or down after that... I also think that if people knew that they got 2 dp for leaving a comment, they would do so more often...those 2 dp add up quickly... |
db0 Level 12
PathmakerPosts: 111 |
Meh, 2dp per comment is next to worthless. People interested in gaining DP just open random pages which is easier and needs less thought. 20dp per comment otoh would be a much better incentive, but i digress. @Verifex
The second option is bound to reduce the overall number of ratings at a time where it seems there are precious few. Personally I'm in for a bigger scale initially, as 5-star systems are really limited, and with clear descriptions of what each rating means like:
Still, with only 5 stars, there is only so much you can do which is why you have 70% of missions in the 3/4 stars rating. |
dux Level 7
PathmakerPosts: 48 |
Whatever rating system we use, if it comes down to many people adding a rating independently, the result will be approximately bell-curved, so most of the ratings will end up in the middle (thanks to the central limit theorem). Whether it's five stars or fifty stars, about 68% of ratings will end up in the middle. It's not due to the granularity of the rating system; it's the shape of the normal distribution. |
|
pixielo Level 20
PathmakerPosts: 3706 |
@dbo 2 dp is not worthless when you've taken almost 2000 missions...
"Take care of the pennies, and the dollars take care of themselves..." |
|
db0 Level 12
PathmakerPosts: 111 |
Well, you run 2000 missions, did you leave 2000 comments? And even if you did, you could have just as well opened a new page. What I'm trying to say is that saying "Leave a comment to get 2dp" is not really any kind of incentive. |
|
pixielo Level 20
PathmakerPosts: 3706 |
I imagine that I left at least 600 comments on the first 1000...maybe a lot more, when there weren't that many missions it was really easy to do...
|
|
verifex Level 10
SeerPosts: 17 |
dbo2:
Which is why I made this post, I didn't make this post to re-examine why the current system is so good it just needs more description. I think the ratings aren't descriptive enough. If each mission had 3 (or more even) qualities that you rated it on, you could only pick one, or two, or all of them, then it would be easier to rate a mission. The more ratings it has, the more descriptive the rating will show. Heck, you could even then let people sort missions by other criteria then what we have right now. |
riscy Level 17 
DestroyerPosts: 1055 |
@dux - what you said, although my knowledge of statistical analysis has long gone, I still remember the bell curve :) |
|
riscy Level 17
DestroyerPosts: 1055 |
@dux - By the way, are you familiar with the 80-20 principle?? In this case 80% of people will take 20% of missions, and the other 80% of missions will be rarely taken (by approx 20% of visitors). |
|
dux Level 7
PathmakerPosts: 48 |
Sure am, riscy. 80% of mission ratings come from 20% of players. |
|
verifex Level 10
SeerPosts: 17 |
If those very 20% of players that make up 80% of the mission ratings had MORE metrics to rate on then just the "overall quality" (which is assumed what the current star system is rating) property then it's more likely we will get something useful out of the ratings. Here's why: While the bell-curve might apply very well to a single property of a thing (in this case a mission), the bell-curve is going to reach it's peak among different collections of properties at different times and for different reasons; making differentiation between large sets of objects (missions) properties (ratings) a much easier prospect. Also, just because 20% of the people make all the ratings doesn't mean those 20% are all going to vote in such a way that a normal distribution is even appropriate. |
|
dux Level 7
PathmakerPosts: 48 |
I think it might be more satisfying and perhaps more interesting to rate missions along a range of critieria, such as (off the top of my head) writing style, thought, thematic unity, interest, etc. I can also imagine having to rate several criteria being a pain sometimes, so it would be good to see a toggle between an overall rating scale and multiple criteria scales. Add together a bunch of bell curves, though, and you get a bigger bell curve. If you want to change the shape of the distribution, you would need to select non-randomly a group of players and give them and only them the ability to rate missions. If you were to search the game database to find players who give missions high or low ratings significantly more often than the general average of players, and gave them and only them the ability to rate missions, you would see the distribution of ratings skewed either high or low. (I'm not advocating actually doing this.) |
|
verifex Level 10
SeerPosts: 17 |
The end user of this data isn't a statistician; Having a big pile of missions with multiple metrics (bell-curve or not) to sort on (sorting is key here) gives the end user (you and me) something to guide us to missions we might like based on more then just one generic criteria (highest rated). (sorry for such a long sentence) |
|
dux Level 7
PathmakerPosts: 48 |
Well, why didn't you say that? ;) |
| Back
