This has been picking at the back of my mind for a bit, since I first saw this thread.
I'm not trying to defend them BTW, just trying to think about the problem in real world terms.
Given the
roughly spherical shape of a differential housing, and how the volume versus height works out, it is the
first portion (~half in your case) of volume that makes the greater change in vertical height of fluid relative to the engagement with the gear set.
Think about it, take a sphere and slice it up (think calculus) in even vertical layers. The very bottom of the sphere has almost zero volume for your first layer, so it's very easy and quick to fill, but the higher you go up in slices, a greater volume of fluid makes a smaller change in height, by the time you get to the center of the sphere, a relatively large volume of fluid makes a relatively small change in height.
(Not relevant to this problem is that one you pass the "equator", the problem reverses and each vertical slice requires less and less fluid for a larger increase in vertical height, when you finallyget to the top and that last bit of sphere requires almost no fluid volume to fill.)
So the point here is that of the two "halves" of your fluid volume, it is the first half of fluid volume that makes the greater vertical fill difference relative to submerging the gearset in oil, the second half makes a relatively small change in vertical height, relative to the gearset, which is what you care about.
Besides the inherent tendency of the ring gear to pick it up and sling it.
NONE of this excuses them not filling it properly BTW!
Check this video out from Gale Banks, just the part of the fluid action, although I like the whole thing, seeing inside and what the fluid really does inside is pretty cool, really showing that the oil goes EVERYWHERE.