@choess we're intending to deviate from POWO here right? Does this describe the deviation you'd like to see?
https://www.inaturalist.org/taxon_framework_relationships/366278

You're right that you could represent https://www.inaturalist.org/taxon_framework_relationships/366278
as a many-to-one with Spinulum, Dendrolycopodium, Diphasiastrum, Lycopodium as Internal Taxa and Lycopodium as the External Taxon and then lots of additional 1-to-1 TFRs for each descendant (e.g. D. dendroideum and L. dendroideum)

But after playing with these for a year, I now strongly prefer to take a more holistic approach with Deviations and try to combine TFRs that deal with the same issue (in this case the Lycopodium split and its downstream implications). I find this a lot more intuitive and it cuts down on the number of Deviations to get your head around.

I'm working on documentation as we speak but a TFR should include all taxa involved in the change that have different names/ranks or mean something different (eg. broadened or narrowed) across the trees.

So here we need to include Spinulum, Dendrolycopodium, Diphasiastrum since they are not represented externally, Lycopodium since it means something different (narrowed on iNat and broadened externally) and all of the descendants with different names (e.g. D. dendroideum and L. dendroideum).

My preference is also to keep TFRs as small as possible to fully contain an issue (e.g. you could in theory make a single deviation for a whole family all the way down to species but it would be insanely complex so better to have several different ones)

Taxa with the same name/rank but different parents ('alternate position') don't need to be included in Deviations (I remain a bit torn on this, but if we include them, mostly because of additional nodes like subfamilies single individual TFRs get unmanageably complex)