These days I've been searching for the state-of-the-art equation discovery system. The goal of these systems is to help human experts discover behaviour patterns in data expressed as equations. We plan to realize an experiment on using metadata to assist search over usually large hypothesis spaces.In the talk I will mention some systems and improvements since 1986, with emphasis on Lagramge, the system chosen for our experiment. This should give an idea of how equation discovery systems work. Some comments on the field challenges to conclude.