One of the main problems with a purely descriptive account of laws is that it renders those laws epistemic.

What does "being epistemic" (or "purely descriptive", for that matter) mean? Taking a photo of a man doesn't render the man "purely photographic", nor is the image "purely photographic"; the image of the world describes the world, and both have the properties reflected in the photo. Predictions inferred from known physical laws coincide with the events in physical world, and this systematic coincidence reflects the presence of common structure between the world and its description. It is this common structure that makes known laws useful, and if it were imprinted on a different description it would remain so, but being shared with the world (and many other things) it doesn't exclusively belong to our descriptions.