UX can easily hide most of the ugliness. Similar URI schemes are in use over many protocols. From the web, to LDAP, ActiveDirectory, and user experiences have been built that hide them, extremely well. Having a URI scheme allows for multiple resolver protocols, which is pretty important if you want to break free of centralization of the Certificate Authority (CA) architecture we are all reliant on, and is very vulnerable to state-level attack.
Discussion
I helped design DID (well, a bit). It's perfectly possible for different systems to have different URI schemes, in fact, it's encouraged at internet scale, which is the path to interoperability
nostr already has decentralized identifiers. In fact they are more decentralized than DIDs because the canonicalization does not depend on an http context, or on external vocabs and schemas, including xml schemas
It would not be hard to give nostr a URI scheme, and I think it can be done by cherry picking the best parts of DID, and throwing out the bloat, the technical debt, external centralized dependencies and stuff that doesnt work, such as range of the did resolution function
nostr URI schemes can be simple, self-contained and truly decentralized, allowing internet scale interop
So far people are coalescing around nostr:npub.. type things, which I wasnt fond of at first, but is growing on me
Plucking concepts from DIDs but not trying to conform to spec – this is what I’ve been arguing for. This is the way.