I think it a grand irony that the only graphs on this page that aren't barely-interpretable depictions of category error are, of all things, the pie charts.
It's so bad; the curves are hurting my eyes... my favorite part is where the curves get clipped to a suddenly straight line. That only seems to happen when the numbers are the min/max of the data range. Whoever wrote this fitting algorithm needs to be drawn, quartered, and fashioned into a pie chart macabre.
It makes no sense to draw a (smoothed!) curve between e.g. 5 releases 450 days ago to 3 releases 420 days ago. The most charitable interpretation is that the y axis is releases/month, but (a) a rolling average would be a more appropriate visualization for that figure (b) the numbers are clearly actually just discrete releases ascribed each of the points at 30-day intervals (c) that's not what the label says anyway.
The other line graphs have basically the same problem.
Meanwhile, pie charts are infamous for being abused to show things that don't actually add up to 1, but in this case they're actually being used fairly appropriately.