For publishers it means that owning and selling content is now becoming a relic of the past, and only still exists because of the considerable inertia in the system. Their role is now clearly what it always was in disguise, namely a service to authors. The need for print and distribution made it perhaps inevitable to get a distorted view of the situation, and selling the carrier – paper – was easily confused with selling the content, but a publisher's real 'market' was, and is, authors. That's also – as appropriate for a 'market' – where the competition takes place. Only authors have a meaningful choice between publishers; libraries (readers) do not.
As for libraries, their role in the past has always included dealing with publishers. After all, they are in charge of the incoming collection of literature, and making sure that their constituency of readers has access to what it needs. Now that the role of publishers has become more clear, namely the role of service providers to authors (which is clearest with open access publishers), many librarians still feel the need to be involved, this time on behalf of their constituency of authors. But authors never were the library's true constituency, at least not when it comes to the authors' dealings with publishers.
So why should libraries take it upon themselves to play the intermediary between authors and publishers. Have authors asked for that? Have university administrators asked for that? Have funders asked them to take up that role? Or is it a consequence of the publishers asking libraries to support OA publishing? I suspect it is the latter. But libraries could decline. They are not in charge of research funding, so why should they be involved in paying for the necessary publication of research results, the cost of which is an integral part of doing research? It certainly isn't provided for in their collection budget.
There is of course nothing against libraries being in charge of the 'outgoing' collection – the papers written by researchers at the library's institution – as well as the incoming one. But it is my impression that a clarity of understanding of such a role is missing, particularly of the budgetary implications.
Take this blog entry in "The Scholarly Kitchen". It is fairly typical for such comments to take as read the idea that libraries pay for OA article processing charges. But logical it isn't. And any comparison between the cost of subscriptions and the cost of OA publishing is bound to be misleading, as it is, in the expression of Stevan Harnad, comparing apples with orangutans. (The analogy may be more appropriate than might first appear: fruit/frugivores – articles/researchers ['textivores'?].)
Library 'membership' of an OA publisher is a supporter-scheme for OA. A stimulus in OA's early stages. It can never be – and shouldn't be – a subscription substitute. Susan Klimley, the Serials and Electronic Resources Librarian in the Health Sciences Library at Columbia University, who is quoted in the blog post, got it right (I paraphrase): The library creating author funds to pay article processing fees helps to reinforce a fundamental disconnect between who creates, and who pays for, article publication. Setting up a pot of money is not going to solve that problem. Authors need to be more sensitized to the cost of producing information, and author publishing funds work against that aim.