I believe the search for the single best model is becoming more and more misguided in a complex world with randomness, measurement error, observer bias, and so on.