I just returned from vacation and saw your detailed post, which
took some work, thanks.
Your Welcome
I think I understand what you are saying, and will try to
reiterate some of the main points, hopefully correctly, to see if I
get it.
ok
On initial flow, the cooling action between the mold and metal
will create an air gap, and thus a volumetric reduction for both
inner and outer diameter surfaces. This form of shrinkage explains
why it is sometimes necessary to create a bulge in a wax when a
flat plane surface is desired in the final casting.
The volumetric shrinkage occurs first thus creating an air gap.
Convexing a surface to allow for volumetric shrinkage was quite
common, and in all honesty Iāve never personally approached a flat
surface in this manner. For me, it was all about material delivery.
But there is also a second form of shrinkage occurring, that which
happens when the metal crystallizes into a complete solid. It is
for this type of shrinkage that inner and outer diameter surfaces
can vary closer to or farther away from the mold walls.
Whatever happened during volumetric shrinkage, will be reflected in
the final casting. Once it becomes a solid, it will be dimensionally
stable for as long as it stays at that temperature. However, as soon
as you quench, what everyone else told you comes into effect and that
is a predictable shrinkage on mass towards the center point of the
casting and not to the center of mass as was the case with the
initial pour.
A combination of surface tension, gravity and orientation produce
a "desire" of the molten metal to reform in a particular shape.
Well you added some points that I hadnāt touched on :-), but close
enough.
Thus parts of the mold will be fighting the metal's force from
cooling, while other parts of the mold will have an increased air
gap and be unstressed by the casting.
On a poorly designed mold as in die casting, this could be the case,
and is something you try to avoid. In lost wax casting not
necessarily
so. If that flask is held at casting temp indefinitely, the air gap
remains. Allow the flask to cool to room temp, then there will be
pressure on the mold and normally it will shrink onto the core.
If only the first type of shrinkage occurred, draft angles would
not be required for release for multi-use die cast molds. But they
are needed, which means that the crystallization shrinkage force is
dominant over the lava-tube like volumetric shrinkage.
Neither is dominant, because at the pre solidification and at
elevated temps, the volumetric shrinkage is occurring. Once the
casting is solid, it will remain stable until it cools below that
solidification temp, then and only then will the global shrinkage
come into effect. Therefore, they are non-competing reactions to keep
it simple, and merely an evolution through process. In die cast
molds,
1 degree of draft will usually do it. However, there are many
instances where you can get away with no draft at all, but in
reality, the mold will last much longer because of less friction
during ejection. Now in sand casting, that draft is critical to allow
the pattern to release efficiently and leave a nice clean impression
in the drag base.
Predicting and compensating for this shrinkage with risers, ribs,
draft angles, radii and all the other tricks you mentioned are
what make casting an industrial science, a profession I now more
fully appreciate (from a respectful distance).
You and me both 
Best Regards.
Neil George
954-572-5829