Bluddy 'ell!
I started reading that thread, and didn't get out of it until about 2AM!

Anyhow .. I've read Marianne's threads on DPR previously, but don't frequent those boards due to the general nature of them.

Very illuminating information on the inner workings of the AF system, and so far she seems to have presented a very good argument for her case.

Two points I find unresolved as yet.
It's still a bit inconclusive that the amount of light at the AF module begins at an f/5.6 aperture equivalent, even tho much data has been posted to prove the idea.
The images captured by snellius appear to not only prove the point, but also to disprove it as well.
(I'm not a DPR member and won't join in that forum every again(unless they seriously consider some level of moderation, so I can't post these questions directly to snellius or Marianne)

The images snellius shows from underneath the AF module definitely indicate that light levels are reduced starting at about f/5.6.
But, the images taken from the front of the camera/lens, don't correlate as well to support the arguments made.

Looking at the images coming from the front, at f/1.4 it's clear that the entire AF module is visible .. most notably the centre cross pattern portion.
At f/2.0, the periphery AF points are still visible, but at f/2.8 they are obscured .. even tho we can clearly see the centre cross pattern AF pattern.

While it appears that Marianne is probably completely correct, the images presented seem to leave a small level of doubt about the actual workings, as opposed to the implied workings.

The second point not touched on is that as the AF modules most important component is a imaging sensor(just like the sensor that captures the image) software allowances haven't been made. This would be either improbable, or impossible to determine and Nikon wouldn't be willing to reveal such info either.
That is, what level of gain does the AF module sensor receive at differing aperture values .. and different lenses mounted.
Considering all this electronic communication between camera and lens, I seriously doubt that Nikon haven't used some software tweaking in the AF system to cater for various lens types!

Anyhow, as you said, the last few replies are going a bit OT(although still pertinent to the OP's original point).

It is interesting and on topic to note that, if true the AF module can only see apertures of f/5.6 or so. I can't imagine Canon's and other brands using completely different PDAF module methods. While they may have differences .. the basic processes will be similar/same.

So that f/2.8 lens may not focus as quickly as it may otherwise do!

I'm happy to continue in a new thread if the interest to do so exists.