If the data is (generally) correct, but formatted badly, imperfect data is better than no data (so it shouldn’t just be rejected). If the data is generally just wrong, the release can always be removed, and it takes only slightly more effort than before (but that’s not that common). If the new add is a duplicate, it should be merged (and the MBID, which is already available in any replicated servers and elsewhere, redirected) rather than rejected (and the MBID lost).
The first and third of these issues were a big part of why the change happened (to avoid people just downvoting things that should be dealt with in a different way). Another reason was that it cuts down on the voting queue in a way that makes actually destructive edits (that can’t just be undone with ease) more visible and likely to get attention.