Suppose for a second I were trying to describe how a Black Labrador Retriever might leave tracks in the sand at a beach. I would begin presenting facts relevant to getting at an answer. I would first say "dogs have 4 legs." I might then lead into other relevant information perhaps about the height labs tend to reach, the speed at which they walk, length of the leg, and by the end I could use this information to put together a pretty good idea as to how the tracks might look in the sand.

A critic of my argument might come to me with a dog who was in a terrible accident, and had to have a leg amputated. "Look, this dog has only 3 legs. How can you claim to know how his tracks would look in the sand?"

And of course he would be right, my claim that one paradigm could truly be a glove tight fit on every instance of a Black Lab would be preposterous. But where does the critique get us in terms of the discourse of the argument? Is there a black lab to which my mapping does not apply? Yes. Does that make my mapping useless to a footprint-identifier determining if some tracks in the sand were those of a black lab? Obviously no.

This is the problem with this type of argument, and one of the reasons I find discussions like this so frustrating. I could use my strategy to create a very accurate depiction of the way Black Labs leave footprints. I can even say without being even a little disingenuous that I have "mapped out the footprint of the Black Labrador." Am I generalizing when I claim that dogs have 4 legs? Of course I am. Is it at the expense of any real attempt at the truth of the matter? Of course not.

The part of it that's frustrating is that at NO point in the argument with my critic did we ever discuss if my mapping of the footprints was RIGHT. There was no investigation of my measurements or sources of the information. We can't even reach the point of discussing the actual attempt at an answer because we're stuck debating in circles about the poor 3 legged animal. At the end of the day, if I convince my critic that my model has weight, it could still be wildly incorrect as to its actual application. If we could move past the exception and get to the meat, we might be able to refine the model to truly be a useful tool for understanding. He could ask to see my ruler and notice I was reading it incorrectly, or find out that what I thought was a Black Lab was instead a Shih Tzu. As it stands, resolving the critique doesn't even let us get any closer to what we really want: an accurate representation of a black lab's footprints in the sand.