I've been engaged recently in a discussion on Twitter as to the appropriateness of Creative Commons licensing for Open Access journals wishing to remove permission barriers, ie. become "libre". It seems there is a real problem as to how to define academic work.
My original instinct was to go with a CC-BY-NC license as this would allow redistribution, use etc. in a context that was non-commercial. This would mean that authors could publish openly at this stage and, if a commercial publisher later wanted to republish (in an anthology for instance), they'd have to negotiate with the author. The problem, though, is that academic work cannot straightforwardly be defined as non-commercial. If a publication comes out with a commercial publisher that makes reference to another piece of work, this could close down citations.
In short, is what we do on a daily basis (reading, writing, researching, publishing) commercial? If we are paid a salary, does that make it commercial? If I publish with an commercial publisher, that almost certainly makes it so. Some of these considerations have already been touched upon by JISC in their paper on this topic.
This is, I would guess, the logic that PLoS went through when deciding their license and I am beginning to come round to that opinion myself, to cite Beckett. While the conversation is still ongoing, it looks as though there is no license that adequately defines non-commercial as "reprinting in entirety, in a commercial publication" and, as a result, CC-BY is the way to go.