Roberts et al., in Building a "GenBank" of the Published Literature  argue compellingly for the following three pleas to publishers and authors: It is imperative to free the refereed literature online. To achieve this goal, the following should be done:
The goal of freeing the refereed literature online is entirely valid, optimal for science and scholarship, attainable, inevitable, and indeed already overdue . But Roberts et al.'s proposed means alas do not look like the fastest or surest way of attaining that goal, particularly as there is a tested and proven alternative means that will attain the very same goal , without asking journals to do anything, and without asking authors to give up anything:
(i) There is no reason journals should pre-emptively agree to give away their own contents online at this time. If researchers wait until many or most journals find a reason for doing so, it will be a very, very long wait. (PubMedCentral has only twenty willing journals so far, out of many thousand refereed biomedical journals).
(ii) Asking authors to choose which journal to submit their research to on the basis of whether or not the journal agrees to give away its contents online for free rather than on the basis authors currently use -- journal quality, reputation, impact factor -- is again an unreasonable thing to ask, and will result in a long, long wait. More important, it is an unnecessary thing to ask, as there is already a means for authors to achieve precisely the same goal immediately without having to give up anything at all: by self-archiving their refereed articles themselves, in interoperable, University Eprint Archives .
(iii) Creating new journals, without track-records, to draw away submissions from the noncompliant established journals, is another long uphill path, and again it is not at all clear why authors should prefer to take that path, renouncing their preferred established journals, when they can have their cake and eat it too (through self-archiving).
In an editorial response to Roberts et al.'s article, entitled "Science's Response: Is a Government Archive the Best Option?" , AAAS has announced itself willing to free its contents one year after publication (see my critique ).
(The New England Journal of Medicine  plans to follow suit, and undoubtedly other journals will soon do so too).
In the service of the same objective as that of Roberts et al., Sequeira et al., in PubMed Central decides to decentralize  announce a new policy from PubMedCentral (PMC). PMC already accepts contents from publishers who are only willing to free them 6-12 months after publication. Now PMC is ready to accept just the metadata from those publishers, linking to their toll-gated websites, if they agree to give away their contents on their own websites 6-12 months after publication.
This is another path that is likely to take a very long time to reach its objective. And even then, can research really be called "free" if it must wait 6-12 months to be released in each instance? Scientists don't rush to make their findings public through PUBLICation in order to have free access to them embargoed for 6-12 months .
Free access to refereed research a year after publication is better then no access, but it's too little, too late. And there is no reason the research community should wait for it. Delayed release is just as inadequate a solution for this anomalous literature -- written by its authors solely for its research impact, not for a share in the access-blocking toll-gate-receipts (for which the majority, royalty/fee-based literature is written) -- as lowered subscription/license tolls are . Lowered tolls, like delayed release, are better than nothing, and welcome in the short-term. But they are neither the long-term solution, nor the optimal one, for research or researchers.
Currently there are six candidate strategies for freeing the refereed research literature:
(1) - (5) all require waiting for policy changes and, even once these are available, all require a needless sacrifice on the part of authors. With (1) the sacrifice is the needless author offprint expense, with (2) it is the author's right to submit to their preferred journals, with (3) it is (as before) the author's potential impact on those potential users who cannot afford even the lowered access tolls, with (4) it is the impact of the all-important first 6-12 months after publication, and with (5) the sacrifice is the quality of the literature itself.
Only (6) asks researchers for no sacrifices at all, and no waiting for any change in journal policy or price. The only delay factor has been authors' own relative sluggishness in just going ahead and doing it! Nevertheless, (6) is well ahead of the other 5 candidates, in terms of the total number of papers thus freed already, thanks to the lead taken by the physicists.
It is high time for all the other disciplines to follow this lead, rather than to wait, contemplating needless sacrifices and nonexistent obstacles. Interoperable archive-creating software is available, free for all universities to install and their researchers to fill . Just go ahead and do it! The details of the self-archiving initiative for freeing the entire refereed corpus now (including questions about copyright and embargo policies) are fully described in Harnad (in prep) . A complete archive of the ongoing discussion of freeing access to the refereed journal literature online is available at the American Scientist September Forum , .
Department of Electronics and Computer Science
University of Southampton Highfield, Southampton