For some time I have been struggling with understanding the value of models. I think I have finally come to resolution.
Here's the dilemma I have struggled with:
Models represent truth: without exaggeration science is about creating models to represent Nature. So clearly models are very important.
Models are junk: I see people creating computer simulation models and then making predictions using these models. Being a programmer, I know that programs can be made to say anything. "Models are junk" I say.
Here's the resolution I have arrived at:
A model must be based upon a theory. The degree to which the theory has been observed to correspond to reality is the degree to which we can have faith in any predictions made by the model.
So, when someone touts a prediction made by a model, ask:
- What theory is the model based on?
- What experiments have been done to see how closely the theory corresponds to reality?
Without good answers to these two questions, any prediction made by the model cannot be trusted.
Have I come to a good resolution of the model issue? Are there other factors I should consider?