[pdonis]

> For what useful definition of deterministic?

The definition that says the future state is entirely determined by the present state. That's the only definition I'm aware of.

>the state evolves in a predictable way from "god's eye", not from the perspective of the experiment occupying any given branch.

The entire "god's eye" state is the one that appears in the dynamical laws of QM (unitary evolution), so that's the one that's relevant for assessing determinism.

> the state evolves in a predictable way from "god's eye", but not from the perspective of the experiment occupying any given branch.

This "apparent randomness" of measurement results is equally true of chaotic classical systems; it's not something that only appears in QM. Basically, it's just a consequence of the fact that individual "observers" will in general not have complete knowledge of the state. That doesn't mean the state doesn't evolve deterministically; it just means the observers don't have complete knowledge.

[photonic29]

>This "apparent randomness" of measurement results is equally true of chaotic classical systems; it's not something that only appears in QM.

Is that a fair comparison? Yes, in either case, the experimenter is limited in his predictive capability by the information available to him. But in a chaotic system, your predictive power can be improved arbitrarily by surveying more information with greater precision. As I understand it-- and hopefully you can clarify if this is accurate-- decoherence forbids a measurement from receiving information from a branched outcome, so even if you take a measurement with arbitrary access to information now and repeat the same measurement in the future, there becomes a set of information that is fundamentally off limits to the observer in a given branch.

[pdonis]

> Is that a fair comparison?

I think so. Perhaps it will help if you look at it this way: you repeat some measurement multiple times, and get a sequence of results that looks random. Is the randomness because of classical chaos, or because of quantum "indeterminacy"? From the measurement results themselves, in many cases, there will be no way to tell. The only case in which there would be a way to tell would be if you specifically made measurements on entangled quantum systems in order to test the Bell inequalities; if those inequalities are violated, the measurements can't be due to classical chaos. But that just underscores my point: looking at "apparent randomness" of measurement results is not sufficient to tell whether they are due to "quantum indeterminacy.

> decoherence forbids a measurement from receiving information from a branched outcome

Once again, this is a misleading way of stating it. What is happening, again, is that the observer evolves into a superposition, corresponding to the superposition that the measured system is in. Decoherence just means the branches of the superposition don't interfere with each other. But the system is still in a single state; the "branches" are not separate states or separate entities, they're parts of a superposition.

(Note, also, that decoherence does not guarantee that the different branches will never interfere with each other. Decoherence is not a fundamental limitation; it's just a recognition of what happens in the usual case, where no special measures are taken to isolate the system or to facilitate interference. According to the MWI, there is in principle always a way to cause the different branches to interfere, i.e., decoherence is never absolute.)

> there becomes a set of information that is fundamentally off limits to the observer in a given branch

According to the MWI, the observers in different branches are not different observers; they're different terms in a superposition that the observer is in. Thinking of them as "different observers" with access to different information implicitly assumes something like the Copenhagen interpretation.