You might as well be saying that if fluid dynamics models have struggled to help predict or explain CPU performance increases, we should have the same level of expectation about how helpful a fluid dynamics based model of aerodynamics will be, i.e. complete nonsense