if you compare the ones that are open (https://www.inaturalist.org/observations?place_id=21&subview=table&taxon_geoprivacy=open&taxon_id=120976), you’ll notice that these are at the species level, and the ones that are obscured are at subspecies.
it used to be that global taxon geoprivacy would take precedence (see https://forum.inaturalist.org/t/question-about-un-obscuring-taxon-geoprivacy/2838/12), but around 2021, that was changed to allow the taxon geoprivacy settings for particular places to override parent place settings (https://forum.inaturalist.org/t/when-there-are-conflicting-conservation-statuses-geoprivacy-obscuration-should-be-prioritised/60358/5), but maybe the change doesn’t work for subspecies taxa?
we can do a test. theoretically, if i add a subspecies-level ID to an observation currently at species level, then it should become obscured if the 2021 change mentioned above doesn’t work for subspecies. (note adding a disagreeing species-level ID to an observation at subspecies level will have no effect because i think the taxon geoprivacy works on the ID taxa rather than the observation taxon, just to be conservative.)
before:
after:
(so then if subplace taxon geoprivacy overrides don’t work for subspecies taxa, you’ll probably need a feature request or bug report to get it changed / fixed. or if you don’t want to wait on a system change, then you could probably remove the global and other parent place setups for the subpecies taxa.)