Hmm, but if the principle comes with a model-dependent length scale, that experimentally looks wrong, why should I discard the principle when it might as well b

Hmm, but if the principle comes with a model-dependent length scale, that experimentally looks wrong, why should I discard the principle when it might as well be an incomplete model? Or frankly inaccurate estimation, my understanding is that it is extremely hard to actually sully simulate the lambda-CDM model from Big Bang.