Suppose, for example, the global cooling following a large volcano matches the cooling simulated by a model, but the model underestimates the magnitude of this cooling: the observed global cooling can still be attributed to that volcano, although the error in magnitude would suggest that details ...