I haven’t yet read David Ellerman’s article, but I wonder if the non-crossing partitions have a special logical significance? Unlike regular partitions and like subsets, they are self-dual as a lattice.