If the models are really wrong in a fundamental way and this can't be fixed by some tweaks in the free parameters (and this is a BIG if), then it's hard to pred

If the models are really wrong in a fundamental way and this can't be fixed by some tweaks in the free parameters (and this is a BIG if), then it's hard to predict what the consequences might be: the model is wrong, in a way that hasn't been extensively studied. How different a correct model might be is hard to predict.