We're talking about two things. You've conflated them incorrectly because you're letting Anders Björkmann teach you his special brand of physics. Capsizing has to do with the balance of roll moments and righting moments and has nothing to do with a change in buoyancy. Sinking has to do with a change in buoyancy, which has nothing to do with roll. In your haste to appear smart, you've cited factors for one as if they were factors for the other, and are frantically trying to suggest that your critics don't understand "simple" physics. And now you're waffling your way out of a practical experiment to demonstrate that "simple physics" doesn't do what you say it does.
Displaced with respect to what? What are the forces involved? Upon what centers do they act? What physical phenomena define these centers? In what directions must each force act? What are the force vectors involved, and how do they sum? Draw us a diagram that shows that a mere change in the magnitude of the buoyancy vector will also necessarily change the center where it acts and thus create a roll moment that inevitably capsizes the ship.
No. When a vessel lists at 90° it lists at 90°. When a vessel lists at 180° it will have capsized belly-up because that's just a different way of saying that's what its roll angle is. This is a function of the vector sum of forces and the dynamics of it are driven by factors that are only partly related to the magnitudes of these vectors. You present a source that deals only in the magnitude of vectors (in fact eliminating the vector reality altogether) and cite it as proof that the directions of these vectors and where they act upon the structure of the ship makes capsizing inevitable.
In practical shipbuilding, once a vessel has rolled to particularly degree and after the righting moment is no longer strong enough to restore trim, openings in the vessel become vulnerable to shipping water into the hull that a were intended to remain above water. The notion that any ship whose "hull is not breached" will float upside down for hours in any roll attitude is naive.
A superstructure changes the vector sum, to be sure. But in ways your model of the dynamics doesn't properly consider. Draw us vector diagrams of a ships with and without a superstructure and show how the sums change as buoyancy increases or decreases.
Ah, we're to the, "You may be an expert, but I say you're lying," stage of the conspiracy-theorist's pattern argument.