Well, as you can see I'm trying to keep up with several people at the same time and hopefully I'm not confusing your arguments with them and thereby answering questions you didn't ask.

But your example of C. difficile is interesting, for I think it actually proves my point.
You claim that they have objectively "low biological fitness" because they have a very low poplulation, especially compared to other bacteria.

This does not actually have anything to do with the gene pool, unless you would say that all species of bacteria share the same gene pool (and besides, bacteria have a different type of gene pool anyway, since they don't sexually reproduce).
None the less, your clasification of "having low biological fitness" is principally based on their low frequency, not based upon their internal genetical code, or their strong phenotype in and of itself, but only on their frequency.

Next you say that when anti-biotics are applied they are the only ones (or one of the very few) who can survive and therefor increase their numbers.
And once antibiotics are stopped things go back to normal.
You conclude: "That is an example of natural selection. It does not necessarily select for any particular characteristic."

Great, but that's my point!
My problem is not with the notion that nature selects, or that the survivers contribute more to the gene pool or anything, but my question is that if NS does not select for "any particular characteristic", how can that knowledge help us understand NS?

Now you can come with a specific example where NS does select for trait X, but the problem is that NS as a whole cannot be "tested". It can be (and if you ask me: must be) assumed, but I can't think of a non-tautologous formulation of NS that is explanatory.