My conclusion: its a disaster and all isced:level keys with values containing 5 or 6 need to be deleted and re-done. See Key:isced:level - OpenStreetMap Wiki to get an inkling of the XXXXXX.
Yes, it’s a mess. isced:level=* has other usability issues besides. ISCED explicitly isn’t designed for classifying individual school facilities. The official correspondence for some regions is too coarse or conflates categories that the general public considers distinct. No official correspondence exists for some subnational school systems. But that didn’t stop some imports from making assumptions that no longer make sense.
I think a data consumer would only be able to salvage the ISCED tags by interpreting it like crossing_ref=* or designation=*, with the values only being meaningful within a given region, except that they’re all integers for some reason. In other words, the project to globally harmonize school classifications for querying purposes is pretty much fantasy at this point.
I favor leaning into regionally relevant classifications that satisfy our verifiability expectations: school=* to record a more literal school classification based on local terminology, grades=* to record grade numbers where applicable, and of course min_age=* and max_age=* for kindergartens and other facilities that admit by age. Granted, it’s still plenty of work for editor developers to create regional presets for these tagging schemes, particularly school=*.
Great, it’s almost as if the proponents of this key want to deprecate it as much as I do. I think the local communities that have invested a lot of energy into this tag may want to do an audit as soon as possible, reinforcing the problematic isced:level=* values with school=* or similar based on inspecting history or conducting a fresh survey. Otherwise these skunked tags will be wasted effort.