All models are wrong, but some are useful.
George E. P. Box, Norman R. Draper (2007). “Response Surfaces, Mixtures, and Ridge Analyses”, p.414, John Wiley & Sons
All models are wrong, but some are useful.