I am pretty sure Marianne's experiment is flawed

You've been thinking too hard.  Time for you to do an experiment.  This is fun:

You may easily test the accuracy of results obtained with a front aperture, by observing its effect on the COC from an out-of-focus point source of light.

You are likely already aware that the shape of the aperture diaphragm appears in such blur circles, with enough detail that you can see the exact shape of the diaphragm, e.g., how many blades it has.

If an aperture placed in front of the lens has a similar ability to project its shape into the blur circles, then it must be equivalent to the lens internal aperture.  Set up your camera pointed at a reasonably bright, tight light source in a dark scene.  Defocus enough to produce a sizable blur circle, then place a sheet of cardboard or foil in front of the lens, with various-shaped holes cut into it, such as squares, triangles, stars, etc.  How well are these shapes projected into the blur circles?  If they are well-defined, how can the front aperture function differently from the internal aperture?

Also see this thread for an interesting example.

--
Qualities possessed by God in infinite proportion: Love, Grace, Power, Righteousness, Wisdom, . . .
Qualities possessed by humans in infinite proportion: Ignorance.
- Marianne