I don't know exactly how this should be designed (some user research seems in order before coming up with any solution). The problem to me is how to design subscription/synchronization mechanisms giving people freedom to choose which data to reuse or not and which "fixes" to send upstream to a centralized knowledge base. I believe this is how the relation between Wikidata and other projects was originally conceived: something like this would allow structured data to be broadly reused without neglecting the very legitimate concerns, policies and expectations of data consumers.