Is this Really Random?

DeviouslyCursed:  I don't think that's it. The pool of players should still be representative of the population, so even though WoT is matching tanks and such, the PR distribution should still be the same.

DeviouslyCursed:  Yeah, the more I think about it, the more certain I am that is has to be #3 of my post. WoT is pulling from a relatively smooth distribution curve, but I'm betting your program is effectively pulling from a set of plateaus. This will have the effect of over representing worse players on the left side of the median for each color, and over representing good players on the right side of the median (assuming skill increasing from left to right). This would most likely have the effect of showing a wider disparity of skill than there actually is (in other words, making the distribution graph wider). Well, after taking a look at Neato's data, I didn't realize yours was that far away from the WoT (hey, those numbers on the bottom mean something!), so maybe the pool of players wouldn't account that much difference. Maybe you just messed something up somewhere.