It may surprise some people to learn that I am supportive of the notion of 'author pays' publishing (or more correctly, of publishing models supported by author-side payments since the actual cash often comes from funding bodies or other sources).
It won't surprise anyone to know that I have some caveats:
Firstly, publishers should be left alone to determine their own business models - those that are best for them and for the communities that they serve - and this means being free of, for example, interference from governments which pushes them in one direction or another.
Secondly, the actual level of fees that journals would need to charge is likely to be very much higher than the typical fees currently seen in the marketplace. My [wholly unscientific, so you'll have to trust me on this] experience would suggest that fees would need to be many thousands of dollars (but more research like that undertaken by the Research Information Network is needed).
Thirdly, it is pretty clear that as things stand adequate funds are not yet available to cover these costs for all disciplines. It may be possible in time to divert funds away from library acquisition budgets to pay author-side publication charges (no, I'm afraid that any monies 'saved' from cancelled journal subscriptions won't get used for all of those books you've been wanting to buy...') but that's not going to happen overnight...
Fourthly, and related to my third point, there is the issue of the differences in funding from one discipline of academic endeavour to the next and the impact that this has on the affordability of 'author pays'. Whilst I admire greatly the clarity of the Wellcome Trust's statement that they see dissemination as part of the research and are therefore willing to pay 1-2 per cent of their research budget to fund that dissemination, that only holds for well-funded subjects where research grants are relatively large. The absolute cost of dissemination is likely to be broadly the same irrespective of discipline but in subjects with modest research grants like mathematics or ecology the percentage cost of dissemination will be much higher than 1 or 2 per cent - perhaps even exceeding 100 per cent. So the dissemination could well cost more than the entire research budget.
Fifth, it is very important to ensure that the mechanism of publishing is accessible to everyone and that means that in an 'author pays' publishing system those that can afford to pay have to subsidize those that cannot (this happens in the subscription world through organizations like INASP).
Sixth, it is not a sin for publishers to make a profit (surplus, call it what you will...). Profit is an incentive for good publishing, results in innovation, and means that publishers have funds available to invest in the future of scholarly communication (journal publishing is a model industry in terms of adoption of new technologies and it is just grossly and ludicrously wrong to suggest that they have been slow to embrace the Internet, for example).
Now, before this turns into a version of Monty Python's 'What have the Romans ever done for us' in reverse, let me explain why I think 'author pays' publishing could be a good thing for some publishers.
A) Benefit to the advancement of knowledge
I don't think you can argue against the fact that Open Access could bring benefits to some disciplines for example if the whole corpus of literature was open to interrogation by data and text mining tools. If you want a compelling case for this, talk to Peter Murray-Rust (or let him talk at you!) and find out about the interesting stuff the chemists are doing...
B) Scalability of income
I've been involved with running a couple of journal publishing businesses in my time and one of the problems we had was that we couldn't raise the price of the journal fast enough to cover the costs of the extra papers we were publishing. We managed to increase the quality, and kept increasing the rejection rates of the journals, but there were still more and more high-quality papers to be published. A model where the income scales more or less directly with the number of papers you are publishing therefore seems like a good thing (I also think that submission, rather than acceptance, fees are an interesting concept but that will have to be the subject of a future post...)
C) Small publishers with no sales force
Again referring to my own experience of running a smallish publishing house, there was no way that I could deploy an international sales force trotting around the globe polishing the doorknobs of librarians and selling our wares. Our relationship with the librarian community was therefore pretty much non-existent. However, we had great visibility among, and relationships with, the authoring community who really appreciated what we were doing for them... so if they paid the bills (as long as they had the money, of course) I would have been quite happy.
D) Eliminates motives for piracy
This, I think, is a big issue and one which is only going to get bigger. In-copyright academic content is increasingly appearing illegally on peer-to-peer file sharing sites and it is virtually impossible for the publishing industry to police this. At the very least it is very expensive. Open access does eliminate the motive for piracy and may, in the very long way, be a pragmatic approach to stopping piracy. I bet the music industry wishes it had the equivalent of an 'author pays' option.
Incidentally, there are a number of less-than-ideal artefacts that are shared by both the subscription and 'author pays' model. In both publishers expend time and money dealing with articles that are rejected and so the higher your rejection rate, the higher your costs. I don't believe that this is an incentive to publish poor articles and the quality of a publication and its brand are going to continue to be very important.