An Instrumentalist interpretation is a description of quantum experiments.
This article aims at proposing a non-interpretative description of some key quantum experiments, based as far as possible on genuine facts, although there is no criterion for deciding whether this goal has been fully achieved or not (should it be feasible).
The "Stern & Gerlach" experiment
A particular "Stern & Gerlach" experimental pattern run in a precise configuration (i.e. with a specific relative orientation between the "green" apparatus and the "red" one) generates a flux of discrete information whose distribution function over a given set of values (e.g. high, low) is stable, reproducible at will, valid for any experimentator. The distribution function is a property of the experimental pattern. Running the experimental pattern results in a measurement of the distribution function. It creates knowledge about a property of the experimental pattern. It adds value if considered within the framework of a theory about experience, about "what it means to be the Actor of an experiment in the world".
The distribution function does not bring any knowledge nor information about "what happens inside the experimental device". It does not add any value if considered within the framework of a physics theory, until it is converted ("interpreted") into a property of our "model" or "simulation" of "how the world is, how it works, what happens there inside the experimental device". Stating that the information gathered through participating into an experiment is "about a system", meaning "about something in the world" is already the result of an interpretation. This statement is neither true nor false: it simply cannot be falsified or verified by an experiment.
For a different orientation of the green apparatus with the red one, the distribution function measured is different, but stable and reproducible as well. And the same is true for any intermediade value of the relative orientation. The distribution function evolves continuously in response to a continuous change of the relative orientation of both apparatuses, it evolves when swapping from one experimental context to another one. This is a fact. In the same way, dis-continuous changes of the distribution function reflect dis-continuous changes of the experimental pattern, e.g. when swapping from an S&G device without shutter to the same one with a shutter. Installing the shutter translates into a dis-continuous change of the distribution function.
When installing a shutter on a specific channel of an "S&G analyser", one component of the state vector gets "filtered", it is eliminated. One cannot conclude that the shutter acts as a physical "filter". All models of what might happen inside the experimental device based on filtering a flux of "particles" fall into contradictions or paradoxes. One cannot either demonstrate that the "state vector" (or whatever it stands for) gets filtered inside the experimental device. The state vector represents formally a property of the experimental pattern. It gets filtered when swapping from a pattern without the shutter to a pattern with the shutter. Changing the experimental pattern translates formally into the filtering of one component of the state vector.
Conversely, the "instrumentalist view" states that the distribution function describes the property of "something" in the world, property which takes a certain value at a precise location inside the experimental device, and evolves from one place to another, e.g. left or right to an apparatus. This instrumentalist view does not report facts. It develops a metaphoric statement resulting from the transcription of experimental facts into a simulation of what might happen in the world. But experimentation will never bring a factual evidence of any property changing value inside the experimental device. Changes of the distribution function can only be evidenced by comparing the outcome of different experiments.
For more complex experimental settings involving several "S&G filters" in a row, with various relative orientations and various combinations of shutters, the quantum formalism allows to compute sequentially the "state vector" associated to various locations, by applying a sequence of "operators" which represent the "evolution of the state vector inside the experimental device from one location to the next". But this is again a metaphoric statement: there is no way to measure the state vector at a certain location, except by placing a set of "detectors" at this location, thereby eliminating all subsequent filters. The quantum formalism is actually modelling evolutions of the "state vector" when swapping from an experimental setting to a new one obtained by adding one more apparatus to the previous setting.
One should not be surprised about the effectiveness of quantum formalism. There is no reason to believe the formalism mysteriously encapsulates any truth about what happens inside the experimental device, something that would need to be revealed via some form of physical interpretation. Quantum formalism simply mimics experimental activity, it explains how we interface with the rest of the world through experiments, it describes the protocol of our interface to the world.