What I think thereisnospork is trying to say is that you can't give the same weight to "discoveries" obtained through simulation using current models as "discoveries" obtained through observations via empirical research.

The models are estimated through the observations we make through experimentation on the real world. They're not necessarily right.