[maxniederhofer]

This reminded me of Jacques Mattheij's list of ideas for startups (see http://jacquesmattheij.com/My+list+of+ideas+for+when+you+are...):

"(45) OpenPapers

A place where all academic research that has been funded in part by public funds is published, journals be damned. Hopefully with deep pockets to fight off the lawsuits."

[ak217]

For NIH-funded studies, PubMed Central is actually doing a pretty good job. It would be great if PMC actually had a mandate to collect papers for all US federal grant-funded research, though.

[jluxenberg]

arXiv seems to fit this bill, and has a reasonable bulk data access policy http://arxiv.org/help/bulk_data