Thanks for writing this. I want to really dig into this paper and make sure I understand it, but it certainly seems like an interesting approach. I'm curious why you say this, though:

Evidently, this approach suffers from a number of limits: the first and the most evident is that it works only in a situation where the system to be observed has already decohered with an environment. It is not applicable to, say, a situation where a detector reads a quantum superposition directly, e.g. in a Stern-Gerlach experiment.

Maybe I'm misunderstanding you, but I thought they addressed this issue:

(from the longer companion paper)

Actually, self-locating uncertainty is generic in quantum measurement. In Everettian quantum mechanics the wave function branches when the system becomes suciently entangled with the environment to produce decoherence. The normal case is one in which the quantum system interacts with an experimental apparatus (cloud chamber, Geiger counter, electron microscope, or what have you) and then the observer sees what the apparatus has recorded. For any realistic room-temperature experimental apparatus, the decoherence time is extremely short: less than 10^20 seconds. Even if a human observer looks at the quantum system directly, the state of the observer's eyeballs will decohere in a comparable time. In contrast, the time it takes a brain to process a thought is measured in tens of milliseconds. No matter what we do, real observers will nd themselves in a situation of self-locating uncertainty (after decoherence, before the measurement outcome has been registered).

As long as there is macroscopic decoherence before the observer has time to register any thoughts, the approach seems to hold, and that's certainly the case for Stern-Gerlach experiments.