First impression from someone who implements lots of Web specs: it reeks of bull crap. I don't see the point in standardising a new URI schema. The example shown at <https://w3c.github.io/did-core/#parts-of-a-did> comes with much expense: verbosity, dereferencing (i.e. network round-trip) to get a document with additional required information and drags in the json-ld spec for no good reason.
If I were in charge of minting identifiers, I would always just use the http URI schema; the controller/verification stuff could be simply appended right there. With just a modicum of care and planning, the URIs will last as long as the Web itself. http://enwp.org/PURL http://w3id.org