Mmm no that's not what I meant.Fredrik, are you trying to make the distinction between model and meaning, where a mathematical model helps us make predictions but does not necessarily give any understanding as to why that model works, as opposed to a theory which tells us why things are as they are? Isn't this the same as the ontological vs. epistemological perspectives? If so, I would be more comfortable if you would use that language. And I think you'd have to put it in one of the philosophy forums.
What I meant is - spawned by the note that any given observation can give rise to a multiplicity of different actions and choices about the status of current theory - that a theory and theoretical frameworks in particular, are MORE than just wrong or wrong. It determines also how new data is treated.
Suppose that a certain amount of evidence "points towards" violation of lorentz symmetry, or supersymmetry.
What is the exact impact of this evidence? This is a decision problem. The problem isn't to kill or corroborate. The problem is how to learn. An theoretical frameworks are more like interaction tools.
When is an interaction tool or learning strategy ever falsified?
Then deeper connection I then make is that; if you consider physical actions of say a proton to be inferencial to it's nature (a generalization of the entropic origin of interactions) then this all this is put in the RG context. The question of how to scale a theory between different observers, becomes that of scaling a decision problem.
All this, suggest that theory and in the extension suggested here, even physical LAW (seen as rationality constraints) are not just descriptions that are just right or wrong.
The extreme implication of this, relating to this context is this: supersymmetry or not may be a matter of perspective! Insist on supersymmetry and you solve some problems but get others; do away with supersymmetry and you solve some yet other problems and created some yet new ones.
The learning strategy here is a CHOICE.