Nice paper, the last two figures remind me of Hierarchical Edge Bundles (2006) and the Sunflower Visual Metaphor (1999) which have similar goals.

https://aviz.fr/wiki/uploads/Teaching2014/bundles_infovis.pd...

https://ics.uci.edu/~kobsa/courses/ICS280/notes/papers/Rose-...

It may be just me, but if the number items in the set gets to numbers that require these sorts of diagrams, then the value of the visualization effectively drops to zero. I mean the purpose of data visualizations is give a human a more intuitive representational understanding of data set. A simple venn diagram does this well. But when you get 50 or 100 data sources, then it just becomes a multi-colored complexity. I mean, they look cool. But I don't gain any better understanding about the data when I look them. It may in fact make it worse because I may misinterpret the size of the areas as being meaningful when they are just necessary for geometry to work.

At least the paper is transparent it's a twist on Edwards cogwheel.

I've encountered these in the wild, so interesting to see what's generating them.

These are hard to read if you have no intuition about what these cog shapes look like even without overlapping. But, everyone get ellipses.

Yeah it doesn't really work in either direction. If I look at a spot its hard to tell which blobs its a part of. If I look at a blob its hard to find all the spots it contains. (imagining of course them not all being labeled) n=4 is already pushing it for venn diagrams, you should probably just have a table above that point.

Impressive but useless. None of there are any more intelligible/useful than circles after 3.