The true costs involved in open access publication (assuming no paper copies) really come down to the labor of the editors and reviewers. This is something that we have been supplying as a free service - whether it be to for-profit publishers or societies (or funding agencies). It is an absolute necessity to have a good peer-review/editing system in place to ensure the integrity of our science and because tenure and promotion systems at most institutions use a researcher's peer-reviewed publication record as a primary evaluation measure.
So, assuming peer-review must be maintained, what is the cost in terms of person-hours to actually review and edit a typical science publication? Back of the envelope, bare minimum, I would say for a publication would involve 2 reviewers X 8 hours + 1 editor X 8 hours, and assume that a manuscript requires revision so double that. BARE MINIMUM then is 48 hours for peer review. What do you want to set as a rate? Let's say $100/hour (it's easy to argue a bit less or a lot more). Feel free to disagree, but I don't believe a peer reviewed open-access publication can be produced for less than $4800. If there are professional editors out there, they will probably tell me I'm way low on that estimate.
Who bears this cost of quality, open-access, science publication is something that needs to be addressed. Currently the majority of the expense is born by the institutions who pay our salaries (if you review papers during work hours) or by our families (if you review your papers on weekends and on family vacations...as many of us do).
side note: For K12 education research, evaluation of materials is a professional service that is explicitly required in many grant proposals. I find it interesting that evaluation of materials through scientific peer-review is assumed to be free, or at least is a hidden cost in grant proposals. One wonders how the latter system developed, anyone know the history of peer-review?