The idea is for “knowledge blocks” to be able to be more granular. Instead of a monolithic document with everything in it, which pushes towards a hierarchical viewpoint, more granular addressing of knowledge blocks through individual files allows data to have one-to-many relationships with related ideas.
I agree this might require a function in DT to view items together, e.g. similar to Scrivener’s “scrivenings” view, which shows selected documents together and allows editing as though they’re the same document.