Skew figures on files incorrect

My understanding is that the skew figures should be percentages of the average. e.g. if you had a file with 20 records on a 10-way cluster the average would be 2 so if theres a node with 5 records that would have a skew of 250, and another node with a single record on it would have a skew of 50. However on a file with 80 parts where they're all on one node we're seeing a min value of 10000 and a max of 790000: