From The Scientist (registration required), we have a few examples of universities trying to increase the accessibility of their research. There are many positions between a traditional subscription based journal and the open access models like PLoS Science, and they all have a place. Brief outlines of some of the alternatives after the jump.
In January the University of California Libraries arranged with the publisher Springer to allow for any article written by a University of California-affiliated author published in a Springer journal to be available immediately to all, even if other articles in that journal are behind a subscription wall. Under this arrangement, UC authors retain copyright and Springer is compensated. It is unclear how articles co-written with non-UC affiliated authors are handled.
Harvard’s arts and science faculty have established an open access archive for publications by its faculty. The library manages the archive and faculty can opt-out of archiving work through a waiver process. Boston University, Griffith University (Australia) and Nottingham University (UK) have established similar systems.
Also worth noting is that the costs of some pre-pay open access journals have proven as budget straining as increasing journal subscription fees.
I don’t pretend to have any clear answers here. I do think maintaining, if not increasing, the availability of scholarship is an important thing for science and for the public. It seems that maintaining a multitude of options would help acheive that goal.