JSTOR is a not-for-profit online archive of over 1,400 academic journals and related content. Many academic institutions subscribe, giving their staff and students online access to a huge body of knowledge. If you don't have access, you can buy individual papers, but the fees are quite high - $14 for On the Method of Theoretical Physics, which I picked arbitrarily.
In September, JSTOR started making older, presumably non-copyright material freely available, and it has recently announced a "Register and Read" scheme giving restricted free access to 70 of the journals it carries, with the expectation that hundreds more will be added. This is welcome news. But the story reporting this in The Chronicle of Higher Education reveals that every year JSTOR "turns away almost 150 million individual attempts to gain access to articles". (This is not JSTOR's fault. It's an archive, not a copyright holder. Access restrictions are the price it pays for being allowed by the journals to carry their content.)
One of the costs of academic copyright is 150 million failed attempts online every year to learn from the corpus of published research.