The "radical Bayesian" stance is very difficult to grasp, coming from an ordinary scientific background. At least it was for me - took months, maybe years of sinking-in to figure out what Fuchs was trying to say. I think a lot of people start reading about it, quickly get the (false) impression that it's solipsistic, and just lose interest.

To be honest, I still can't understand how it's not solipsistic. If you use QBism from two different perspectives, like Alice and Bob in a Bell test experiment, then you'll still either need many worlds or get incompatible results.

Perhaps it would help to note that in QBism, reality is not denied. Rather, the objective reality of the quantum state is denied.

Not sure what you mean by incompatible results. Certainly QBism is incompatible with Many Worlds! The latter states that the quantum state is real and objective, the former states the opposite.

The QBist horsemen have devoted some attention to the question of compatible quantum state assignments. See: https://arxiv.org/abs/quant-ph/0206110

However, Chris Fuchs does not see any problem in someone assigning a probability-1 belief to a measurement outcome and being proven wrong. See section 2.2 of https://arxiv.org/pdf/1705.03483.pdf

This last point in particular should bring home the point that Fuchs does think there is an ontological theory underlying physics, but that we are fooling ourselves about how much of it is objective/observer-independent.