Ok, I guess we need a numerical example here.
Lets say you had two votes. One by Rudy Reg and one by Larry Luser.
Rudy Reg is a respected member who has a 5 rating.
Larry Luser is an annoying newbie poseur and has a 1.
Rudy and Larry both rate you. Rudy respects your contribution to the website and rates you a 5 (congratulations for earning the respect of a respected peer).
Larry is a schmuck and rates you a 1 for the hell of it.
Now, if UBB just used a straight average, (1 + 5 ) / 2 = 3
Now that wouldn't be very fair, would it. Why should a Luser influence your rating like that.
Therefore, UBB uses a Weighted Method. If you want to look at the way UBB handles it, check out ubb_profile.cgi
Yeah, its fairly complex, but the bottom line is UBB will calculate the weighted average for each class of ratees (1's, 2's, 3's, 4's, 5's), then add them together to determine your weighted avg.
Not, sure of the exact calculations, but it is probably something similar to this.
Rudy is a 5 and rates you a 5
Larry is a 1 and rates you a 1
Rudy's 5 weights 5:1 to Larry's 1. So your average is around 4.33.
And yes cal, there is a lot of overhead here.
Personally, I dislike ratings (and no not because I don't rate well), but because I don't have an inferior complex that needs to feed off of my rating

-Tacks