I feel this - I’m often on the other end working with data from clinicians in the field for massive studies. The forms that come in can have an infinite number of possibilities just for noting sex. Enough so that our semantic layer needs a human reviewer because we keep finding new ways field clinicians have of noting this. Now imagine that over the whole gamut of identifiers.
tl:dr - Humans are almost always the problem in data harmonization.
That is deodorant antiperspirant actually prevents you from sweating in the first place