I think the arxiv / pre-print workaround could have so much more potential if more researchers were motivated/educated on it. I'd love to see an effort to simply email the top 10k authors to request preprint copies to be hosted. If all they have to do is send a pdf, I think most would be happy to do it.