Should probably be a log scale for the y-axis – you probably want to know when you’re paying twice as much as you should, rather than 25sats more. In particular an estimate of 150 sat/vb vs 75 sat/vb (the blue spike at the end) is probably about equally bad as a 50 sat/vb vs 25 sat/vb (the yellow overestimate for a while at the beginning).