"What makes for a good model?" In various forms, this question is a question that, undoubtedly, many people, businesses, and institutions ponder with regards to their particular domain of modeling. One particular domain that is wrestling with this question is the multidisciplinary field of environmental modeling. Examples of environmental models range from models of contaminated ground water flow to the economic impact of natural disasters, such as earthquakes. One of the distinguishing claims of the field is the relevancy of environmental modeling to policy and environment-related decision-making in general. A pervasive view by both scientists and decision-makers is that a "good" model is one that is an accurate predictor. Thus, determining whether a model is "accurate" or "correct" is done by comparing model output to empirical observations. The expected outcome of this process, usually referred to as "validation" or "ground truthing," is a stamp on the model in question of "valid" or "not valid" that serves to indicate whether or not the model will be reliable before it is put into service in a decision-making context. In this paper, I begin by elaborating on the prevailing view of model validation and why this view must change. Drawing from concepts coming out of the studies of science and technology, I go on to propose a contextual view of validity that can overcome the problems associated with "ground truthing" models as an indicator of model goodness. The problem of how we talk about and determine model validity has much to do about how we perceive the utility of environmental models. In the remainder of the paper, I argue that we should adopt ideas of pragmatism in judging what makes for a good model and, in turn, developing good models. From such a perspective of model goodness, good environmental models should facilitate communication, convey—not bury or "eliminate"—uncertainties, and, thus, afford the active building of consensus decisions, instead of promoting passive or self-righteous decisions.
