QBism and causes of EPR correlation I wish to clarify my understanding of the QBist (eg  Fuchs et al), or Quantum Bayesianism, interpretation of quantum mechanics and in particular its approach to the EPR problem.
Suppose we consider a canonical EPR-Bohm experiment with two observers, Alice and Bob, measuring the spin of two entangled particles in space-like separated observations.  QBism requires that Alice can make her own observations but must rely on reports from Bob on what he observes.
In this way, I presume that QBism allows Alice to become aware of correlations between her observations and those of Bob. Given a suitable set of experiments, these correlations (or anti-correlations) may be very strong and repeatable.
My question is:- How does the QBist interpretation understand these correlations? Are they simply unexplained correlations without cause, or are they presumed to have a cause. If the latter, how is the cause embedded in the quantum formalism in the QBist approach?
To clarify, the observations in the experiment described above are space-like separated and my question is motivated by this statement in Fuchs et al:-

Quantum mechanics, in the QBist interpretation, cannot assign
correlations, spooky or otherwise, to space-like separated events,
since they cannot be experienced by any single agent. Quantum
mechanics is thus explicitly local in the QBist interpretation.

 A: I see no clear explanation in Fuchs et al of what QBism predicts, but on page 2 they say:

The notorious “collapse of the wave-function” is nothing but the updating of an agent’s state assignment on the basis of her experience. Acting as an agent, Alice can use the formalism of quantum mechanics to model any physical system external to herself. QBism directs her to treat all such external systems on the same footing, [...] even agents other than Alice.

Based on that, I'm going to hazard a guess that what QBism means in concrete terms is the following.
Suppose Alice prepares a qubit in the state $\sqrt\frac13\,|0\rangle + \sqrt\frac23\,|1\rangle$, then measures it in the $\{|0\rangle,|1\rangle\}$ basis. If there are many worlds, then the result is a pure state looking like $\sqrt\frac13\,|00^A\rangle + \sqrt\frac23\,|11^A\rangle$, where $|0^A\rangle$ and $|1^A\rangle$ represent the state of Alice's brain after seeing the outcomes $0$ and $1$. On the other hand, if there is only one world, you end up in a state that can be modeled by a mixed state like $\frac13 |00^A\rangle\langle00^A| + \frac23 |11^A\rangle\langle11^A|$, where the $\frac13$ and $\frac23$ are classical probabilities (or behave like them).
The pure and mixed outcomes are theoretically different (there are measurements that distinguish them), but indistinguishable in practice (it's impossible to make those measurements), unless Alice's brain is a thermodynamically reversible quantum computer, in which case it may be possible. Because we aren't quantum computers, we can pick either one without fear of getting wrong answers, and people do.
Many-worlds is the idea that the pure state is the correct one and the mixed state only gives right answers by accident. One-world models effectively say that the mixed state is the correct one and the pure state only gives right answers by accident.
QBism seems to be the idea that the mixed state is the correct state for Alice to use, while any other "agent", such as Bob, should use the pure state. It's consistent with experiment because we can't do the experiments that might contradict it, but it would be just as consistent to do it the other way around. It makes no difference in QM whether you "had the experience". There isn't any concept of "you" or "agent" in quantum mechanics that could distinguish between Alice and Bob in this scenario.
If Alice and Bob are quantum computers and can do the distinguishing experiments, then things are different. In that case, Alice and Bob need to use the same state – the correct one, whatever it is – or they will get wrong answers for certain
experiments.
... Unless, I suppose, QBism is actually correct. In that case, Alice finds that her measurements do collapse the wave function, while everyone else's do not. Bob appears to her to see the same experiments and agree that only Alice collapses the wave function – but that isn't actually Bob, but a p-zombie. The real Bob is in a different subjective reality, where everyone including the fake Alice agrees that Bob is the only one who collapses the wave function.
Quantum mechanics doesn't work that way, but I suppose we can't be sure that the real world doesn't work that way until we figure out how to transfer our minds into quantum computers.
