Good point. You use this “visible-by-default” field for a valuable piece of information. Logically, it is an attribute but this avoids the pain of creating one through a dialog and numerous clicks.
Personally, I never store Sosa or other such numbering because it is not “absolute”. It depends on the root. Since I use my data to generate multiple ascendant/descendant trees, this would require to recompute the full labelling each time.
Side remark: how do you handle lateral branches? Say your first cousin would be 4AB? And what about non-connected people?
Could you explain the scheme in Families?
In Notes, the ID seems to take part of its encoding from note type. I guess the media ID is a shortened form of repository+description.