Research output is in the end one main criterias that we measure. Actual quality of it not so much. Thus science is driven to deliver publications. Now how valid these are is open to questions. And how many of them actually prove the mechanisms behind the findings? Probably comparatively fewer to those describing some correlation.