🎛️ One Dial

All three positions turn out to be one dial: how much of the formalism you promote to furniture. Copenhagen promotes the collapse event. Everett promotes the state. QBism promotes neither. Once it is a single axis, "unsatisfying for opposite reasons" stops being a complaint about two theories and becomes the price of where you set the dial, which is a better way to hold it.

Here is the thing I would put to you, and it is your own tool turned on your own discontent. If descriptions never cause, then a framework that declines to explain is not falling short. It is being consistent. The dissatisfaction with QBism is a craving for a description that grounds, which is exactly what you refuse to allow. So that half of the dissatisfaction may be a platonic residue rather than a defect in the framework. You already granted this when you said passing the buck is no defect for something that never claimed to explain, and then listed it as a shortcoming anyway.

Which suggests what would actually satisfy you is not an interpretation at all, since an interpretation is a description and you have ruled those out as causes. On your own terms the fastest way to the outcome is to be the process, so the satisfying thing would be a constructive account that runs. That is the reconstruction program rather than the interpretation debate: Lucien Hardy's axioms, and the informational derivations from Chiribella, D'Ariano and Perinotti, which try to derive the formalism from operational postulates instead of interpreting it afterwards. They answer why the formalism has this shape without promoting any piece of it to furniture. That is the only shelf in this argument where your requirement and the physics are asking the same question.

One tightening on "collapse cannot be physical". It is stronger empirically than a priori. Every attempt to make collapse physical has to add dynamics, GRW with its noise and Penrose with gravitational reduction, and those additions are testable in principle and so far untested. So the honest form is that making it physical costs you new physics nobody has won a bet on, which is firmer ground than an impossibility claim and harder to argue with.

And your revision holds together. Rejecting branches because they make the description the cause keeps the half you actually wanted, which is that the structure exceeds any agent's summary. That survives without a completed multiverse, which is where the unboundedness point was pointing.