azaroth42
commented
19 hours ago Update documentation to clarify that if nationality vs ethnicity is not known then it is quite acceptable to not have a metatype at all, and that different organizations can and will make different decisions.
This is a bit provocative, but as I felt I had to create an issue specifically for the problems I have noted while investigating bias here (sorry I should probably paste bit of my comment here but I'm missing time now).
The issue manifests itself further when looking at LUX nationalities at https://lux.collections.yale.edu/view/results/concepts?q=%7B%22AND%22%3A%5B%7B%22classification%22%3A%7B%22name%22%3A%22nationality%22%7D%7D%5D%7D
There are plenty concepts used there, which would probably better fit the ethnicity as it is defined at https://linked.art/model/actor/#ethnicity. For example "Blood" may be seen as a nationality ("Refers to the artwork of the North American Indians belonging to Blood tribe of the Blackfoot.") but probably if rather fit the level above ("Blackfoot"). (btw @azaroth42 I'm using this link as the one I could find quickest, but that pages seems to mix the tribe and the body fluid!) "Aragonese" was a nationality once but it's not anymore, and its description in terms of "style and culture [...] bound by common ethnicity and language" matches better ethnicity. And certainly "Parisian" has never match the notion of nationality. I reckon "nationality" is mentioned in the original AAT concept. But that feels like wrong vocabulary design in the first place.
It's not that I would oppose using AAT's weakly defined nationalities in Linked Art, but for the sake of interoperability it doesn't feel right to use them in a framework of nationality and ethnicity that is presented with much more precise (and exclusive) notions.