Chuck,
I think you're missing the point. Pat's (and I'm certain Chris') point is
that the term "model" used in the way they were discussing, is a
mathematical object that may be built up over things in the real world,
for example, such a model could contain a set containing you and I. So,
standardly, you assign to terms in your language interpretations that are
picked from parts of that mathematical object. Say the term 'funny' is
assigned the interpretation that is the set of you and I. Now, all things
being equal, in FOL if we apply the term 'funny' to John Sowa (the real
John Sowa, not an abstraction of any kind) we get a big fat FALSE. As
Chris and Pat have pointed out far better than I could hope to, there just
is no abstraction going on here of any kind. If you're thinking that the
"model" is about 'funny' and not about sets containing you, I, or John
Sowa, then I could see how you might make that mistake -- things like
'funny' are the kinds of things that make their way into data models and
ontologies. But that is not what Chris and Pat were talking about.