This is very interesting data, and it produces very interesting maps. I was wondering what would be the best projection in this context, and I believe an equal-area projection could be most useful because it would allow us to compare population against area. If we also want an azimuthal projection, which I think is a good choice, then "LambertAzimuthal" would be the projection to use, perhaps with a centering at the center of one of these circles, like the orthographic map above.

I checked the code used in that article some months ago, and it builds confidence on reasonable validity of my solution; it's certainly a separate, and different implementation for the search, yet produces basically the same result. Both rely on effectively the same dataset, so it better be good... :)

Yes, Mollweide is also a good choice, indeed. Thanks for the new maps!