If there is just one label, this is easy enough. But in case that the user is looking for language es, and we have labels in zh, fa, ro and de, how should the software know which one to use?

My suggestion would be to use the following algorithm:

• "direct" fallbacks and variants (what we currently do)
• site content language (we do that currently, maybe we want to stop doing this)
• languages using the same script as the requested one (maybe ULS knows this), in order of the number of speakers (again see if ULS has this info)
• other language by the number of speakers (be prepared to see a lot of chinese)

I'm on the fence about the "languages using the same script as the requested one" is the best path forward. For example, if a Russian user tries to get a label that has Ukranian and English, which should be preferable? I would argue that in a general case, Latin-based scripts have a greater universal appeal (at least in Russian schools Latin script was studied very early on), but I don't know the situation in other countries.

Wondering if there's differents between T13267 and this?

@Liuxinyu970226 related - T13267 allows users to specify a fallback list. This task is about using that list if available, or using a default list for all users.

Before we agree about the correct order of languages, may I suggest we display at least one language label (no matter which?) You can go alphabetically if you like, but something is definitely better than nothing...