Sufficiently well thought through models exist already. Musicbrainz (*) and Wikidata, as pointed out here by @Fernando_Pereira - and both offer a community and (some) curation. Both also come with the promise that if one decides to contribute things will not be lost should a business decide so.
The critical metadata complainer mass is probably too small for Roonlabs to act on it anyway.
As others pointed out that task goes probably far beyond what Roonlabs could deliver. Another reason for me to not contribute to such a solution would be that it would be closed source. The great advantage of adding data to - for instance - musicbrainz is that I can always get my work added to file tags too. There are enough tools out there which make it easy.
And that’s something Roonlabs should act on - creating stunning stuff without the need to have access to the most accurate metadata ever. Make f-ing use of what’s there already. (It might even inspire people to add / enhance metadata at one of the places which already collect it.)
(*) … and it’s not worth discussing that they still add composers to albums at places you wouldn’t want them. Because it’s not an issue if the data that matters is available too. 