Here’s an example where a change based on genetics seems generally unhelpful and having no relation to morphology by their own admission:
The 3 goldenrods (the gigantea/canadensis/altissima complex) formerly together in Triplinerviae are very abundant and very tricky to distinguish, so having them together in the same subsection was really useful for identification. Identifying them up to section Unilaterales means that the identification could also to refer to like 40 other species in 4 different subsections.