Just thinking: maybe we should make it so whoever submits a file can't rank it themselves? Sure if people were shrewd enough they'd go out of their way to make another account and rank from there, but for the most part, it'd be fair...what do you think?

Oh, and two thoughts that immediately came to mind after posting:

1.0 Should it be possible to see what the file's ranking is besides seeing how high it is on the chain, since it's really vague that way?

2.0 Do you get trash everytime someone edits (appends, heh heh) a post? Would you recommend further use of the Preview button, or is that not a big deal anymore? Just for clarity.

http://fateofio.org/project/v4pm/howto/

http://fateofio.org/news/22

Also there's the results of me testing the bracket commands in v5.0

Bug: Commenting on a non-recent file causes crazy problems with the file list. They're corrected easily by updating the ranks, but whoa. O.o

Fixed!

No, and it never did. I mean... with v4, I only received messages when errors occurred. With v5 I don't actually receive any sort of messages. I probably should rig something up to send me exceptions, though...

Oh. Well, the problem is, there is no meaningful way to calculate an "overall rating". A simple average doesn't make sense, because everyone has a different scale. For instance, Dev and I might both think a piece of concept art is equally good, but I might rate it 90 while he rates it 75, just because he's more stingy. So what good would it do to display the average score of 83?

Thus, the ratings people assign are only usable for determining which submissions are better or worse that others.

And, again, rankings are not based on average ratings. Rankings are created by first creating a ranked list for each user (based on that user's ratings), then combining those lists using Condorcet's method.

In fact, as an example, imagine the submissions A, B, and C are rated as follows:

Dev: A 10, B 5, C (not rated)
Temporal: A (not rated), B 60, C 50

Clearly, Dev and Temporal are using wildly different rating systems. In reality, the differences would be more subtle, but there would still be differences.

Now, from the above, the average ratings would be:
A 10
B 33
C 50

Ordering by average rating, we'd have C as #1, B as #2, and A as #3.0

However, if you actually look at the ratings, you see that Dev thinks A is better than B, and Temporal thinks B is better than C. Clearly the correct order in this case, given no other information, would be A #1, B #2, and C #3.0

My algorithm will produce the correct order.

Right. The only thing that matters is which submissions users think are superior or inferior to which other submissions, not the acutal numerical ratings they give.