I am still a huge fan of open science, despite the barrage of pay-to-play spam, and would love to live in a world where all scholarly publications were freely available to anyone who wants to see them. But I am beginning to think we need peer review for journals, as much as we need it for the articles within them.
For theoretical physicists, ambulance chasing involves getting papers out quickly after a major data release. Some ambulance chasers make significant contributions, some are just trying to draw attention to their earlier work, while others are banging out insubstantial papers in the hope that they will be cited by their slower colleagues. But whatever their motives, cosmologists have certainly been busy: the BICEP2 discovery paper has been cited 188 times on the Arxiv, all in "preprints" written within a month of the original announcement. I am pretty sure this is a world record, and you can always check the current tally.
So what gives? So far as I can tell, Elsevier hopes to negotiate blanket deals with science funding agencies and consortia of institutions to cover the cost of these journals. And I suspect many scientists will be apprehensive at the thought of Elsevier inserting themselves even more deeply into the world's scholarly infrastructure.
more importantly, the Archive has reached the point where it threatens to do to traditional journals what MP3s did to record shops, as it represents a radically new model for scientific publishing. In particle physics and astrophysics, the Archive is essentially complete -- I almost never see traditionally published papers that are not also posted to the Archive.