Open access to scientific publications and datasets is very important for both scientific progress and for the incorporation of scientific knowledge into the public sphere. The reasons for this are manifold, and have been enumerated so vociferously on various blogs, email lists, and internet fora that one fears the spittle will be transmitted electronically to accumulate on bystanders' keyboards across the world. Therefore, I'll restrain myself from adding to the ranting here.
Nonetheless, the arguments for open access to scientific products seem to have been effective, and science is slowly opening up. The American Geophysical Union (AGU), which is probably the most important geoscientific society and puts out many top journals (Journal of Geophysical Research, Tectonics, Geophysical Research Letters, etc.) has recently announced an open-access policy, where publications from 1997 to the 24 months from the present (rolling) are open. This is currently 17-ish years and counting of pretty great science, and while it'd be a lot cooler if the entire catalog was open, it's enough that interested parties without institutional access should be able to get up to speed with current research on most topics in geoscience.
Additionally, the Seismological Society of America has a (new?) open access policy that was briefly mentioned in a speech at the 2014 SSA meeting in Anchorage last week. Unfortunately not a lot of specifics were given, but what the SSA's website shows is that authors may post copies of their articles on their own websites simpy by appending a cover sheet to the pdfs. SSA also has open access to SSA members, which I think should be standard for scientific societies, even if whatever circumstances don't allow for complete public access. The Geological Society of America has a similar policy, I believe.