One of the major problem is that bibliometrics criteria are necessarily impacted by inertia of prestige, and publishers own the title of the most prestigious journals.
Older, tenured scientists could just go and stop publishing in those journals to build up the prestige of new (really) open access ones, but the fact is that most often they co-authors papers with their students or at least younger, un-tenured scientists, and they understandably don't want to sacrifice the careers of the new generation.
Some funding agencies now requires that work they fund are made open access, but sadly, publishers lobby in the US and in Bruxelles weight a lot in favor of "authors pay" model, and many funding agencies includes publishing budget in research grant for this reason. That is very frustrating because people mix up real open access with open access paid by authors, while 1- other economical models are possible (see my other comments on this page), and 2- countries and funding agencies could legally enforce that every papers published in a non-open access journal must be made available immediately in an open repository such as arXiv, overruling the decision of publishers who often put an embargo of e.g. 2 years before this can be done.