(Apologies for writing in English, but since an overseas perspective was requested…)
There are minimalist and maximalist schools of thought regarding the localized name keys. From the mailing list debate a decade ago, I noticed that some countries like Germany are stricter about these keys, whereas others like the U.S. are laxer. This is partly due to real-world cultural differences: in a country with an official national language, other languages may seem foreign and less important. Moreover, in OSM, the level of tolerance for “default” values varies from country to country, also when it comes to other open-ended tagging schemes like access tags and payment methods.
When someone recently went around filling in missing names of a few major U.S. cities, many of them identical to the English names, it was largely unremarkable. The only problem was that they uncritically copied from Wikipedia, neglecting to trim off the disambiguators that Wikipedia only uses for technical reasons. If a German speaker had more carefully tagged Phoenix as “Phoenix” in German (not “Phönix” or “Dattelpalmen”), no one would have had a problem with that. On the other hand, if they had blanketed all of Phoenix’s shops and streets with name:de=* tags identical to name=* tags, someone probably would’ve told them they’re wasting their time.
Where would we draw the line? Not based on a string comparison but rather based on notability. How likely is it that this type of feature, with this degree of fame, would ever intentionally have an endonym or exonym in some other language? As soon as a unique name appears in one language, it almost opens the floodgates to all the other languages. People generally seem to be OK with it.
A better example: I have a friend who speaks German, Spanish, and English, in that order. He can certainly cope with seeing London labeled as “Londres”, but he knows that London is “London” in German. A mix of German and Spanish on the map would be jarring to him. German doesn’t have a unique exonym for London, but this doesn’t mean German text would delegate the name to an arbitrary language also spoken by the user.
If I understand correctly, you’re suggesting that, before a data consumer automatically falls back from the user’s first language to their second, it should first check if the first language would use the same name as the local language. Where would it get this information? Perhaps it would somehow know that name=* is in English and that German likes to keep English names verbatim. Such a heuristic would apply globally, but maybe this is not a sure bet for every European language? Some data consumers do take this approach, but only because they’re tailored for a handful of languages or a certain region. I don’t know of a general implementation across all geographies and languages. It would be a big project for linguists.
Alternatively, a data consumer could look at Wikidata, see the German label there, and skip all the other OSM name tags. In practice, that’s what a number of data consumers already do, such as any map powered by Mapbox Streets or OpenMapTiles. But some mappers dislike this end run around OSM. Mappers like the idea of OSM as a self-contained source of toponyms. Unfortunately that doesn’t seem to be compatible with tagging minimalism.