Daniel Allington has written the best thing I’ve yet read anywhere on open access, called “On Open Access, and Why It’s Not the Answer.”
Anyone interested in the question should read it now. It is much more deep and detailed than most of the pro-OA writing out there, and gets at some of the deep political and academic problems that lurk around the stark moralistic rhetoric that informs most discussions of the topic. I won’t try to summarize it, but here are a couple of choice quotations:
“If you do not have access to an adequately funded library, then that is a problem” but it is a different problem from the apparent over-pricing of some academic journals, and requires a different solution. And if you have access to what would seem to be an adequately funded library, but cannot obtain the reasonably priced journal you need because the funds have all been soaked up by overpriced journals, that’s again a problem, but it’s hardly the fault of the reasonably priced journal, its academic editors and contributors, its editorial staff, or even (in many cases) its publisher (since not every organisation that publishes a cheap journal also publishes a very expensive one).”
“Requests for articles uploaded to an institutional repository … primarily come from people who already (for the most part) have access to the same articles through the inter-library loan system.”
“It is hard to see any particular need for an ‘academic spring’: the name by which the most recent phase of the open access movement was, somewhat offensively, referred by some journalists (the implication being that boycotting Elsevier is somehow akin to risking one’s life protesting against a military dictatorship in the Middle East). Completely free journals already existed, albeit that many of them were and are of comparatively low status. There were at least two viable systems whereby people could access articles published in closed journals to which they lacked direct access, namely repositories and (for those lucky enough to be placed within participating institutions) the inter-library loan system.”
“The existence of the larger market enables more money to be spent on marketing (for many academic presses, ‘marketing’ consists of no more than listing a title in a catalogue and mailing out a scant handful of copies to the reviews editors of scholarly journals) and facilitates these books’ appearance on the shelves of general interest bookshops – not because the manager of (say) the local branch of Waterstones necessarily has a commitment to disseminating scientific knowledge (although in practice, that is not unlikely to be the case), but because members of the public are likely to pick them up and buy them, contributing not only to the dividend paid to Waterstones shareholders, but also to the local branch balance sheet, permitting the branch to stay open and the staff to be paid. The irony is that if the text of those books had been published not through the commercial system, but by being uploaded to a “free” website such as this one, far fewer people would have read that text, because far fewer people would have had a stake in ensuring that it would reach an audience.”
“The open access movement is consumerist, i.e. … it has typically ignored production issues and failed to give serious consideration to the academic publishing industry, to the contribution it makes, and to the likely results if it were to be starved of income. This point is obscured by focus … on academics as producers. Such focus misconstrues the relationship that professional academics have with the publishers of academic journals.”
“I am yet to be convinced that social media can provide an adequate medium for the assessment of extended theoretical arguments or in-depth analyses of data, as opposed to strikingly-expressed position statements and technically-impressive visualisations.”
“Although academics collectively appear to be no better than other social groups in considering the welfare of those who depend upon them – Brienza has been perhaps unique in arguing that academics have an ethical responsibility to consider the fate of the thousands of people ‘who make a modest living supporting the publication of worthy research’ – the attitude Fyfe draws attention to is probably more closely associated with new media companies than with open access advocates”

Open Access
I wrote a comment on Allington’s blog to emphasize and expand upon a couple of his key points; it’s reproduced here:
I agree with Sara and will go further: this is the best thing I’ve ever read on the topic. It summarizes quite a few points I’ve been meaning to make in writing about it myself, and I’m grateful I can point people here instead.
Just to chime in, I’ll add a few other points that I believe complement yours:
1) Publicity. One of the main functions of publishers is to publicize (as the name suggests) our work. This costs money. it is one of the reasons Representations and Critical Inquiry count so much in English: they are well-publicized as well as being well-funded in other ways. Along with sustainability, this is one of the things funded publishers can and do offer that completely gratis operations can’t and don’t.
2) Discipline Specificity. The blanket injunctions regarding OA completely overlook the tremendous differences in costs from one discipline to another. As Hal Abelson said in the Swartz report from MIT, the entire JSTOR back and current catalog (mostly humanities and social sciences journals from hundreds of publishers) costs less than the current journal subscription from some individual science publishers (such as Elsevier). If a main part of the pro-OA argument is the cost of journals, then it must take into account the fact that journal costs are radically different across disciplines, as you suggest.
3) Emotion/moral argument. I have my own explanations for where this comes from and why it occurs, a question you aren’t really asking, but I think it has to do with the resistance on the part of OA advocates to really think deeply about what they are doing. There is cognitive dissonance. OA is easily seen, especially in the humanities because of #2 above, and visible in quite a bit of the rhetoric you quote, as a rejection of humanistic academic practice, not as support for it. Keeping these contradictions in mind is hard and produces extra emotion. Arguing that academics and their support system should be paid nothing while simultaneously suggesting you are supporting them is hard work, because it does not make sense on the surface.
4) Mandates. You get near this a couple of times, but there is a tremendous contradiction in the fact that for the first time I’m aware of in history, under the banner of “open” and “free,” academics are being told where and how they can and should publish and not publish. There is a broad suggestion around that you touch on, that academics should not publish in non-OA journals. Whatever the moral benefits of OA may be (and they are much thinner than advocates suggest, as you rightly point out), academic freedom is more important.
5) Libraries. The sharp edge of the OA knife is a Manichean distinction between “open” and “closed.” Anything not freely available on the web is “closed.” This is an amazing reinterpretation of the function of libraries, which have until now been seen as open institutions that provide largely free access to all sorts of published material, and still do. The fact that an article is available for a fee on the web, but for free in nearby libraries, still makes it count as “closed.” That disparages libraries (and is partly responsible for another anti-intellectual push toward putting them out of business) and turns the facts of the world upside-down. If the price of access is a trip to the local library, I don’t see why that is unreasonable. At all.
6) The Reinterpretation of Publishing. In history, publishing has been about making things open and available, through not just printing but publicity, distribution, editing, and so on. Now we have book historians as wise as Darnton reinterpreting publishing itself as a means of preventing rather than providing access. That is really bizarre. “Paywalls” do not prevent access. Stephen King fans are not “prevented” from reading his books because they cost money. This just turns obvious facts on their head.
7) Access for the Disadvantaged. Many publishers and distributors have robust programs to deal with this. JSTOR (again, not a publisher but a distributor), in particular, provides free access to nearly its entire set of journals to almost every African institution and many institutions in developing nations worldwide. (The African program was in place prior to Swartz’s actions, making his mention of African nations in the Guerrilla Open Access Manifesto particularly curious).
8) The Slippery Target of OA. The best arguments for OA focus on academic journal articles because they have traditionally been contributed without compensation. Yet many of the most rabid OA supporters go much further, beyond the Budapest OAI recommendations, and start to talk about mandated OA for all sorts of other things up to and including “everything professors publish.” The fervor with which this position is sometimes recommended (see: the recent AHA Electronic Thesis controversy) also smacks to me of cognitive dissonance, because depriving professors of the opportunity to earn money for their own creative and scholarly productions is one of the best ways to eviscerate what is left of the professiorate.