|
Alright, first, here's how ratings should be handled:
1) User A submits item.
2) Users B-K review said item as follows:
B - 1
C - 10
D - 7
E - 8
F - 8
G - 6
H - 7
I - 5
J - 6
K - 9
3) The lowest 5% of ratings, and the highest 5% are ignored as follows:
B - 1 (low 5%)
C - 10 (high 5%)
D - 7
E - 8
F - 8
G - 6
H - 7
I - 5 (low 5%)
J - 6
K - 9 (high 5%)
4) The remaining reviews are averaged together to define the adjusted rating.
You could even take it further if you wanted, allowing users to vote on individual reviews (rather than reviewing reviewers). In other words, do something like amazon.com does with their reviews. Whenever you see one, you can say whether or not it was useful or valid. The percentage of people who find the review useful is the weight that review has in the overall adjusted rating.
Anyway, just something to think about. Throwing out the extremes will do a lot on its own. It's actually common in many official statistical studies to do this. It accounts for the 10% of the population that are weird, or something.
-----
Next suggestion: This may already be the way it's done, but I'm not sure. The top 10/20/whatever downloads should only take the past week or couple of weeks into account. Who really cares if a file has 10000 downloads if none of them were within the past month?
-----
When submitting articles, there should be a broader range of categories, and users should be able to sort/search based on these categories, or other fields, like auther, date, rating, reviews, etc.
-----
The screen shot idea is absolutely wonderful. I'd like to see shots of both levels and tilesets. At any rate, I think they should be limited in file size and have suggested dimentions (i.e. 160x120).
-----
Being able to submit links would be nice, but just remember who's going to get all of the complaints when the older links start dying.
-----
If you need any mild help with anything, let me know. The majority of the work I do now is on the web, and PHP is basically all I use.
-Onag
|