Necessary Thing

[robvalue]

We seem to be talking entirely at cross purposes.

I'm saying we have an observable set of objects A. Let set B be the unobservable ones.

B might be empty, but we can't assume it is.

Objects in set A may be contingent on objects in group B.

If they are, we have no way to examine this relationship whatsoever. A case as simple as B containing two objects, both contingent on each other, and everything in set A being contingent on one of them, is enough to produce a result where nothing is non-contingent.

We can't simply say "so and so is impossible within group B because of a parallel with group A".

This is exactly why we have the scientific method in the first place, to make sure our models actually match reality. And it's why speculation such as this tells us nothing about reality.

http://youtu.be/inw1fNItjdU