Here's a proposal for solving the problems of access to scientific literature while retaining the integrity of the peer review process:
The International Research Review Authority (IRRA), a non-profit organisation responsible for peer review of the world's scientific literature, would receive submissions from authors accompanied by a $1000 fee, paid by the authors' institution.
The submitted paper would be sent to 5 reviewers, who would each receive $100 on completion of a comprehensive review. The paper would be given a rating by each reviewer on a scale of 1 to 10. Any paper that was rated above 6 would be accepted by the IRRA and therefore indexed in MEDLINE (the 'journal' field would contain 'IRRA' followed by the rating, eg IRRA8). Before indexing, irrespective of the rating given, the author could take back the paper for corrections and further work in an attempt to improve its rating (resubmission being accompanied by a further review fee).
All papers accepted by the IRRA would be stored as XML files in a central archive that would be linked to from PubMed and other central subject indexes. Each paper would be assigned a DOI to facilitate long-term access. No access charge would be made for these files.
Copyright would remain with the authors, who would be encouraged to use a Creative Commons licence for their work. The most likely licence would allow publishers to licence the work for display in their journals. The XML files could be freely reformatted according to each journal's style of presentation, as long as the content remained unaltered. In the case of commercial journals, authors could receive a fee from publishers as a reward for high quality work.
How would this benefit authors?
The current process of submitting papers to numerous journals of decending prestige for peer review would be streamlined into a single submission.
High-quality peer review would be encouraged by rewarding the work of reviewers.
Copyright would be retained by the authors, allowing them to use their work for teaching.
All papers reviewed by the IRRA would be freely and easily available, so research would be based on a full knowledge of previous work and literature reviews would be more complete.
Fees for licencing work for republication could cover the costs of submission and more.
How would this benefit publishers?
There will be no need to carry out peer review. By being able to choose from work that has already been certified, all journals will be able to include papers of the highest quality.
The market for niche journals will open up, as the cost of subscription will be lower.
How would this benefit institutions?
Libraries would no longer have to pay for access to the results of publicly-funded work.
Libraries could carry out their own filtering procedures rather than relying on commercial publishers.
How would this benefit end users?
Central indexes would be able to use the freely available data to increase searching efficiency.
Access to the scientific literature would be free.