On Fri, 1 Mar 2002, Thomas J. Walker wrote: > I agree with Stevan's interpretation that self-archiving of the > APS-formatted version is not legal. > I also agree with his conclusion that APS policy encourages self-archiving > of preprints and updates of preprints.
With so much agreement between my comrade-at-arms, Tom Walker, and myself, including agreement on the optimality and inevitability of open access, how is it that we still disagree on the very same basic points that initiated this Forum in way back in September 1998? http://www.ecs.soton.ac.uk/~harnad/Hypermail/Amsci/0000.html Yet here we are, 4 years wiser, and still not of the same mind on the subject! Raising the question of this divergence in connection with the APS is still more interesting, for the APS too is of more than one mind on the matter: The APS, among all established publishers, has gone the farthest in trying to serve the interests of research and researchers. They have allowed and blessed author self-archiving of both the unrefereed preprint and the refereed postprint (i.e., the final, refereed, revised, accepted draft), and have only been ambivalent about the PDF, the APS-generated page images, worrying that if they allow authors to self-archive that too, it might do in their subscription/license revenue, and make them unable to cover their costs. Moreover, among the costs that the collapse of subscription/license revenue would cease to be able to cover would be the costs of the editing and markup that generate the PDF itself! So it really could be quite risky to encourage authors to do that. On the other hand, not allowing the authors to do that (but allowing them to self-archive the preprint and the postprint) allows a certain empirical, economic test to take place, one whose outcome cannot be guessed in advance: If the refereed, revised, accepted final drafts (postprints) are freely accessible online (through author/institution self-archiving of eprints), but the publisher's PDFs continue to have to be paid for, will that be enough to sustain the subscription/license market, and hence to cover all costs (including the cost of the peer review, which is about $500 per paper in the case of the APS)? If the answer is yes, then the PDF and enhancements indeed represent an essential added value, as many journal publishers have been suggesting, one that those institutions that can afford it are willing to keep on paying for even when the vanilla postprints are openly accessible to all their own institutional users, as well as to the users at all other institutions that cannot afford the value-added edition. This would mean that there is no further reason to expect that the journal subscription/license market will not survive and prosper well on into the PostGutenberg era. [And I, for one, would be perfectly happy with that outcome: It would no longer be true that any researcher with access to the Web lacked access to any of the 2 million annual articles in the planet's 20,000 annual peer-reviewed journals. It would no longer be true that any researcher had lost even an epsilon of the potential uptake, usage, citations and impact of his research because potential users had been denied access because their institutions could not afford the access tolls. And although available serials budgets would still continue to be spent, the situation could hardly continue to be called a "crisis."] But the other possible outcome would be that, as the 2 million annual articles in the 20,000 annual journals all become openly accessible to all would-be users, institutions will no longer find the value-added PDF worth paying for any more, and there will be growing subscription/license cancellations. Then the outcome would be that publishers would have to cut costs, phasing out the add-ons for which there was no longer a market, and downsizing to the provision of the essential service they would still be providing, namely, peer review (i.e., exactly what it is that transforms the preprint into the postprint). That special increment of added value is something that the APS is currently allowing to be given away for free by its authors, even though it is parasitic on the subscription/license revenues with which it is competing! For it is those revenues that are currently paying all the costs -- the essential ones as well as enhanced ones. Downsizing and phasing out everything that goes into the PDF would not only mean that refereed journal publishers would become peer-review service providers (and certifiers), but it would also mean that their "clients" were no longer the reader-institutions but the author-institutions, and they would have to start charging the $200-$500 it takes to generate a refereed, accepted, certified postprint to those author-institutions -- which, fortunately, they will have more than enough windfall revenue to pay for, as those institutions will also be the reader-institutions who were now making all the annual subscription/license cancellation savings. Now what role might Tom Walker's author-purchased PDF offprints have in all this? First, right now, it would confuse the issue a little, asking authors to pay for the PDF out of the same funds they would have used to pay for paper offprints (and for the same price). It is unclear what percentage of the authors of the 2 million annual papers in the 20,000 annual journals fit this model at all. (How many had the $200 to pay for offprints before, and want to spend it that way now? Many never had it.) But let us suppose that authors do have the $200-$500 for this. Now we have two parallel empirical/economic tests: As before, will institutions that were paying through subscription/license tolls for the paper and PDF want to continue paying for it, even when the vanilla postprint is accessible online for free for everyone? Will authors who were paying for paper offprints want to continue paying for them as PDFprints, even when they can make their self-archived vanilla postprint accessible for free for all -- for free for themselves? And an interaction: Will institutions want to keep paying the basic subscription/license fees when BOTH the vanilla postprints and the PDFprints are accessible free by other routes? (The paper add-on is a further complication here, but let's ignore it.) Tom thinks that author-payment for the PDFprint is a temporary tide-over method for the eventual transition to open access. But written out in longhand like this, it looks more like an author-end reprieve for PDF, and a delay in the natural processes that would have to take place in the transition to a self-sustaining cost-recovery model for open access. And it would be the PDF that would be artificially propped up by it all, at the author-end making it more difficult to arrive at the outcome of the empirical experiment of whether there would be a reader-end market for the PDF at all, if access to the vanilla postprint were free! > However, for a society that so clearly takes member interests into account, > I cannot understand why APS does not allow its authors to pay a fair price > to have the PDF versions of their articles freely available concurrent with > paper publication. If APS wants to serve its members and the cause of free > access, it should offer an immediate-free-web-access (IFWA) service. Such > a service would include posting the PDF versions on arXiv and allowing > authors to self-archive their APS formatted versions on any server. The effect of this would be to offload (some of) the PDF costs onto authors and leave subscription/license expenditures in place: A very odd sort of subsidy from authors who could just as easily self-archive their vanilla postprints instead, for free. > Offering an IFWA service would have minimal direct costs to APS, so a fair > but profit-making price should not exceed 75% of the cost of 100 paper > reprints (e.g., $269 x 0.75 = $202 for a 7-page article). > > The Entomological Society of America has offered an IFWA service for two > years, and in 2001 had a gross income of $31,259 from IFWA sales. [For > IFWA, ESA charges 75% of the cost of 100 paper reprints, but its prices for > paper reprints are less than half those of APS. ESA gets $95 for IFWA to a > 7-page article.] The real question is: What is the total revenue per article, how much of that is to pay for peer review, and how much to pay for paper/PDF/etc.? For the average article in the 20,000 journals, it is about $2000. So if $500 of that is peer review costs, what is the other $1500 paying for? (And what on earth does the $200 per article PDFprint charge cover, and what difference does it make?) I gave average figures. Perhaps for the ESA journals and the APS journals the figures are slightly different, but you see my drift. The arithmetic does not work, as a strategy for making the transition to open access, paid for at the author-institution end. > Here are some advantages that would accrue if APS offered an IFWA service: > > APS authors could have the official version of their refereed articles > immediately and conveniently available to all. "Official" here merely means the PDF version. As far as science is concerned, the essential version is the refereed postprint. > APS authors could avoid preparing corrigenda for their arXiv'd preprints. Well, yes, there's that... > Users of arXiv could avoid having to combine preprints with their > corrigenda [possible only if authors post corrigenda]. True again.... > And of course APS would have a revenue source that would help it transition > to universal free access. And what would those further transitional steps be, then? Stevan Harnad