there must be work on the problem of modeling in relation to science. but all parties to the debate had better start by acknowledging that computer modeling is not the same thing as empirical research into the actual world. it is empirical investigation, we might say, into a fabricated or virtual world. now i am not saying that modeling could have no value in science, but the question of the relation of the fabricated to the actual world has to arise as it does not, for example, when we're carbon dating a fossil. this is not to say that questions of the relation of representation to reality (let's say, of a hubble telescope image to a nebula it depicts) don't constantly arise in science of all kinds. but they arise in a different way or at a different level in computer modeling the earth's atmosphere. here we are faced with the specter of losing our grip on reality by the methods we've generated to investigate it: it would be easy to be seduced by the elegance or interestingness of some model and to traffic exclusively in representations. it is no doubt striking when the model predicts apocalypse, and more striking the more apocalyptic that apocalypse is, and how you deal with that is a challenge to your personality as well as your research. it would be interesting to realize somewhere down the line not that reality is really virtual or something but that we lost our grip on reality and became practitioners of a virtual and scholastic science.