By the way, I am not trolling either. I am trying to find out what's left unsaid, when a certain model is recognized as a good representation of reality.

So, we have a model. No matter how accurate it is, there are some free variables that are meant to represent the initial conditions. Usually these are adjusted to best fit the data that we have - that is, minimizing the residuals (the difference between what the model says and what is observed). When analysing residuals, there is a statistical test to see they are white noise. The p-value at which we accept it as white noise represents the uncertainty of model's predictions.

Is this process similar to the best models we have? Do you know if there is clarity (or "consensus of scientists") on what that uncertainty is?