(Attention Conservation Notice: this post assumes some familiarity with Open Access. I’ve included some background links but the essential introduction to OA is by Peter Suber: see his Open Access Overview and also the one-page version thereof.)
In his latest blog entry, Jan Velterop takes an entertaining but (I think) overstated swipe at Stevan Harnad for the latter’s postion on recent moves by publishers to get OA mandates to encompass the paid version, that is, where publishers offer OA on any manuscript in exchange for an extra fee. Harnad argues, if I read him right, that this amounts to a land grab by the publishers, since if their efforts were to succeed the governments in question would be mandating further payments to publishers, on top of the fees and subscriptions they already charge. Since authors can already self-archive, pace Velterop, this really is a kind of alms for the publishers. Velterop takes issue with what he sees as dismissal of the role of publishers in providing peer review and authentication:

Those with a ‘harnadian’ inclination should really not bother publishers at all with their articles. They should just ‘archive’ (read ‘publish’) them in some repository and move on. Shame the articles can’t be labelled as having been published in a peer-reviewed journal, which would make them more valuable and be noticed and taken seriously, but hey, everybody can see them and the publishers just haven’t been able to beg enough cash to publish them.

I think this is a mis-statement of Harnad’s postion in re: peer review. Neither he, nor any serious OA advocate that I know of, has ever indicated that they do not recognise the value in peer review. Harnad’s point, in the post in question, is not about peer review at all but about co-opting a government mandate to direct research funds into publisher’s pockets, when a free alternative is available.
That said, I think there’s more to mandatory OA to publicly funded research (of which, let me state up front, I am strongly in favour) than a simple choice between for-pay or free models. TANSTAAFL. Velterop:

Subscriptions, on the whole, currently sustain the journal system. But they have a downside. They do not, by definition, provide open access. So that’s why new publishing models have emerged that do.
Unfortunately, Stevan derisorily calls these new publishing models PPA, for ‘Paid Publisher-Archiving’. As if ‘archiving’ is what publishers do. Nobody pays a publisher for archiving and no publisher asks for payment for archiving. Publishers ask for payment for having an article peer-reviewed and formally published in a reputable journal.

Well, yes, but remember that researchers provide a good part of the value of peer review — the actual reviews! — to publishers for nothing. Publishers are therefore charging money for co-ordinating the review process and for “formal publishing”, which I take to include archiving and distribution. This is a legitimate charge for legitimate added value, but there is a limit to what the market will bear, particularly as costly print versions of journals move closer and closer to obsolescence. (I haven’t picked up a print journal in years, excepting only those old publications which have not yet been digitized.) At some point it could conceivably become attractive for researchers — perhaps through the NIH, or professional societies, for instance — to co-ordinate the review process themselves, construct a robust search architecture that encompasses the vast majority of institutional repositories and thumb their noses once and for all at the STM publishing industry. (Indeed, when it comes to constructing a global virtual archive, the Open Archives Initiative has done much of the heavy lifting already.) So, let the publishing companies beware. Researchers, if I know my tribe, don’t want the hassle (and we are talking about a mind-bendingly HUGE endeavour, if we are considering moving virtually ALL peer reviewing out of the current commercial infrastructure) — but if pushed far enough, they’ll act.
Harnad and other OA advocates make much of the fact that there is currently no evidence that OA reduces journal subscriptions. I think this may be somewhat disingenuous. Evidence drawn from physics/arxiv is of limited value in making predictions about the much larger and more lucrative (or, from a subscriber’s point of view, expensive) biomed field. Further, if the subscription model still has a much greater market share than any OA based model, it may not mean much if current subscription levels are not (yet) falling as OA grows. Harnad himself has written about the transition from the current system to universal OA:

An alternative outcome [to 100% self-archiving] is that when the refereed literature is accessible online for free, users will prefer the free version (as so many physicists already do). Journal revenues will then shrink and institutional savings grow, until journals eventually have to scale down to providing only the essentials (the quality-control service), with the rest (paper version, online PDF version, other ‘added values’) sold as options.

This is the outcome that strikes me as likely: who is going to subscribe to a journal whose contents can be had for free? I have been saying since the days of E-biomed that OA would mean a significant reduction in the size, both physical and financial, of the STM publishing industry. Don’t get me wrong, I shed no tears over the likely loss of a few companies and shrinkage of the rest; but the important question is how to maintain the integrity of the peer review and authentication system. Will journals (that is, the companies that publish them) simply accept the new order of things and quietly “downsize” until they provide basically just the quality control function? Harnad writes:

In none of these outcomes [see here] is peer-review itself compromised or put at risk; nor do authors have to give up, even temporarily, submitting to their established journals of choice.

Er, well, it may not be quite that simple. If I put my Doom&Gloom hat on, I can see a number of “established journals of choice” simply going out of business, leaving open the questions of what happens to their archives and, more importantly, how do we replace the resulting hole in the quality control infrastructure? We certainly do not want to allow a power vacuum into which will rush the remaining publishers, likely the biggest and some of the worst, gleefully wielding a new near-monopoly. Harnad writes:

Self-archiving could be done virtually overnight.

It could, but it won’t. Getting researchers to work together, even for their own good, is worse than herding kittens. We (the research community, and open access advocates in particular) are NOT going to wave a magical open-access wand and present the publishers with a fait accompli in which they must quietly acquiesce. Government mandates of OA to publicly funded research will go a long way towards forcing their hand, though — which, to return to the original point of this post, is why Velterop is mostly wrong and Harnad is mostly right: researchers should come out strongly against the attempt to have such mandates include what Harnad calls PPA.