question: how to obtain those? We don't have access to every pypi mirror in the known universe. And over what time-period?
and the original premise was: 'most-downloaded list' does not necessarily translate into 'best', or even 'understandable and useful guidance for new site admins and integrators', but that is another discussion.
That of course can mean that having two (or more) lists is a better:
most downloaded
subjective, hand-curated list
so if you have tips on how to generate that most-downloaded one, that's most welcome.
@keul this initial list is the result of @sally's work with the Jazkarta team: they did evaluate a bunch of add-ons and this was their recommendation. So, yes, this is partly a result of Paragon.
The process in the github repo is not voting. Voting becomes a popularity contest, and as Paul wrote, that is not what we are aiming for.
Download statistics does not tell us whether an add-on is well maintained, whether it has tests, whether it has been translated, etc. It would be far more work to try to create the infrastructure to get meaningful stats than it is to get a human curated list up.
If someone would like to implement "phone home" (opt in?) feature that tells us which add-ons are actively used, that would be nice, but that's a separate issue. Ploud also tried to provide error statistics in add-ons (among other things).
TL;DR: please file issues in the repo and we will try to move this forward.