I think in part this is because a lot of people started to be interested in the meta-metrics of scientific work and like any metric once you start tracking it you influence the system you are tracking. Publishers and various scientific actors then made things worse by making those very metrics (a symptom) a goal in its own right. That's what broke the system.