Handshake is awful. Because they decided to allow anyone to register a TLD there's no way to determine if a name needs to be resolved with handshake other than trying to resolve it.
The Handshake design still seems like a step in the right direction, but even with sound data, it's probably years away (if ever) for DNS resolution to "flip the switch" to use this kind of architecture.
There are, however, other reasons why one might think Handshake isn't up to par just yet. Those reasons might be, for example, even though it completes the internet, making it significantly more secure and sovereign, the current implementations including DANE aren't as fast. Of course, this too [1] will be addressed with time.
There are highly technical people working on these problems with a heavy lean on making sure the experience is comparable if not better than today's expected experience. I think things will only get better from here.
Seems like you might be caught on the infinite abstraction escalator though, as who will manage the DNSNS (Domain Name Service NameSpaces) servers, and who'll take the onus for making the standard selector or display for the user to decide which lookup to consider canon for the request? This would actually be a pretty exciting development.
This would greatly complicate domain squatting (or maybe intensify/arms race it), and add a layer of resilience in the form of building up the interfaces required for users to maintain their own private DNS namespaces. This would conclusively shut the door in terms of centralized seizability of domains, since the user would in effect be empowered to be their own registrar.
You could even keep the traditional stuff seizable I guess. However, you'd be able to ask the fellow down the street if he still had the old DNS entry for x.y.z