Maklak brought up a good point about our method of normalizing skills based on ranks.

He noted that our method was plotting skewed data in a tight group, and a large section at another end of the distribution.

I figured a way to resolve this by segmenting the data from one side of the median, and transforming both sides separately.

This is specifically for skills and preferences. It will normalize any values >Median as 50 to 100%, whereas those that <=Median, will be kept just shy of 50% by factoring them up a little bit.

Pic of it in action
http://imgur.com/rls1joX

already been verified with larger data and it approaches .5 Mean.