I thank D. Harry for bringing further attention to the questions of numerical modeling goals, strategies, and approaches, and for clarifying that simulation models—those designed to reproduce the behavior of specific natural systems as accurately as possible—can serve valuable scientific purposes. I agree with this point and allude to it in Murray [2003], a fuller version of the earlier discussion in Eos.

In the situations Harry describes, we believe we know the main physical and/or chemical interactions that are important in a system. In such cases, simulation models can be useful for addressing more detailed questions, as Harry outlines, to help improve our understanding of particular natural examples in various ways.