Your comment makes no sense. Just because there's modeling involved that does not make it a hard science. A hard science requires stuff like the ability to perform controlled experiments and replicability, in order to arrive at a high degree of accuracy in predictions.
Throwing around partial differential equations does not turn something into a hard science. You need to meet way more requirements before you're in a position to claim that.
The scientific mindset is there, but not the publishing because the publishing destroys the value of the model. Once your model is known, others trade against your model and it becomes invalid.