I thought PLOS (http://www.plosone.org) was created to be a disruptive agent
Somebody needs to come along and start by making the little, obvious things right. Their next goal is to avoid acquisition long enough to make a difference. Every so often a promising business comes along, only to be bought out by the big guys and slowly dismantled (see SerialsSolutions).
Also, science and math journals are only an aspect of journal publishing, albeit the most pricey of subjects of which to build a collection. These subjects have always been the first to lean towards open access but they stand in distinction from the rest of a library's collection.
They are no lightweight in the biomedical research journal landscape.
One of the interesting things about scientific publishing is that the current arrangement is a recent phenomenon. Many journals started out as independent or university-affiliated publications which charged to cover costs; then were gradually spun off or bought out by for-profit institutions through the 60s and 70s. The accretion of publishers into a small number of companies, each with a virtual monopoly on critical journals in the field, resulted in the insane balance of power we currently experience. Inelastic demand for information (it's extremely difficult to do research without access to the major journals in your field) means Elsevier et al can jack up prices almost arbitrarily.
Besides, it's not just PLoS , there's Frontiers (frontiersin.org) and many other open access publishers. In fact many paywall-journals allow authors to make their papers open access by paying an additional fee.