JSTOR
does distribute the public-domain papers free of charge. It's the post-1923 papers, which are copyrighted and which JSTOR doesn't own the copyright to, which are paywalled. I do think they could be more aggressively pro-access, but they're slowly moving in the right direction.
That's not to say there can't be other scanning projects that aim to do a superior job, and I'd probably volunteer if there were a way I could be useful to such a project (I've spent some time at Distributed Proofreaders). But I don't see JSTOR as exceptionally evil, at least any more than Google Books is. Both are bringing more content online, in ways that are partly good and partly flawed.