To elaborate on my yesterday's simplified and more humoristic comment: I really think we don't need to choose between a database and a tagging library.
While a database with the sole purpose of tagging your files can be more simple, having a more elaborate database that captures more details of the music world does not harm tagging and can even improve it.
The discussions we have seen that bring up this "tagging vs. database" argument are IMHO mostly about personal preferences in how data should be represented. Often it is about standardization vs. "as printed on the release", and the standardization is seen for the pro database side of discussion, the "as printed on the release" for the pro tagging library. But there is no rule saying for tagging I want everything as written on release.
There are many choices to be made: artist as credited or standardized, artist in native script or transliterated into Latin script, something like "featuring" standardized or not, capitalization etc. etc. Even whether typos should be corrected or not can be a preference. Those choices are not unique to tagging but can be made for other uses of the database as well.
In the end dumbing down MB to be merely a tagging library solves none of the above and doesn't improve the tagging situation either. But building a comprehensive database, that tries to capture the often messy reality of music as good as possible, will ideally allow the user of such a database to make the choices outlined above.
So for me MusicBrainz should clearly continue to become that comprehensive music database. And taggers should make use of the data by providing their users the choices to customize the tagging.