> A baseball is an "inert" object
> that doesn't know anything about any other objects, and yet it seems to
> respond properly with everything else in its environment.

On the contrary.  Both the 'baseball' and the 'everything else' here are
macroscopic generalisations of particular configurations of basic particles
[whatever they eentually turn out to be, but let's call them Atoms for now].
Atoms 'know' about electroweak theory, strong nuclear force, gravity, and
quantum physics --- the first three of which are ways in which atoms *have*
to know about other atoms, potentially some distance away.  These systems
are certainly not inert.

> I think it might interesting to consider an object oriented system that
> models the "inert" object approach for all objects.

Model at a low enough level and you can do this.  But your model rapidly
outstrips available computing power.  Or you abstract a notion of 'baseball'
out as a geeralisation, at which point you also have to abstract out all
sorts of notions about how the class of baseballs interact with the class of
humans.

		- Peter