7.1. Researchers: Self-archive all present, future (& past) papers

This is why it is hoped that (with the help of the eprints.org institutional archive-creating software) distributed, institution-based self-archiving, as a powerful and natural complement to central, discipline-based self-archiving, will now broaden and accelerate the self-archiving initiative, putting us all over the top at last, with the entire distributed corpus integrated by the glue of interoperability
(http://www.openarchives.org/).

As to the past (retrospective) literature: The preprint+corrigenda strategy will not work there, but as the retrospective journal literature brings virtually no revenue, most publishers will agree to author self-archiving after a sufficient period (6 months to 2 years) has elapsed. Moreover, for the really old literature, it is not clear that on-line self-archiving was covered by the old copyright agreements at all.

And if all else fails for the retrospective literature, a variant of the Preprint+corrigenda strategy will still work: Simply do a revised 2nd edition! Update the references, rearrange the text (and add more text and data if you wish). For the record, the enhanced draft can be accompanied by a "de-corrigenda" file, stating which of the enhancements were not in the published version.

(And of course the starting point for the revised, enhanced 2nd edition, if you no longer have the digital text in your word processor, can be scanned and OCR'd from the journal; by thus distributing it, authors can do for their own work for-free what JSTOR http://www.jstor.org// is only able to do for the work of others for-fee.)

7.2. Universities: Install Eprint Archives, mandate them; help in author start-up

For researchers who profess to be too busy, tired, old, or inexpert to self-archive their papers for themselves, a modest start-up budget to pay library experts or students to do it for them would be a small amount of money very well-invested. It will only be needed to get the first wave over the top; from then on, the momentum from the enhanced access and impact will maintain itself, and self-archiving will become as standard a practise as email.

But what needs energetic initial promotion and support is the first wave. If (i) the enhanced access of their own researchers to the research of others and (ii) the enhanced visibility (Lawrence 2001a, 2001b) and the resulting enhanced impact of their own research on the research of others are not incentive enough for universities to promote and support the self-archiving initiative energetically, they should also consider that it will be an investment in (iii) a potential solution to their serials crisis and the possible recovery of 70-90% of their annual serials (toll-access) budget.

(Note that the success of the self-archiving initiative is predicated on the same Golden Rule on which both refereeing and research themselves are predicated: If we all do our own part for one another, we all benefit from it. Give in order to receive...)

7.3. Libraries: Maintain the University Eprint archives; help in author start-up

Libraries can also facilitate a stable transition through their collective, consortial power ( SPARC : http://www.arl.org/sparc), providing leveraged support for publishers who are prepared to commit themselves to a scheduled for downsizing to the essentials only (the peer review service, to the author/institution). And individually they can also be preparing in advance for the restructuring that will come if their windfall toll savings grow; about 10-30% of their annual savings will need to be redirected to cover their university's own authors' peer-review charges per paper. The remaining 70-90% is theirs to use in any way they see fit!

7.4. Students: Stay the course! Surf! The future is optimal, inevitable and yours!

7.5. Publishers: Concede graciously on self-archiving and be prepared to separate essential peer-review service costs (to the author-institution) from optional add-on product costs (to the reader-institution)

A much better policy is to concede on the optimal and inevitable for research, and plan on the possibility of separating the provision of the essential peer review service to the author-institution (peer review implementation charges, per paper) from the provision of all other add-on products (e.g., on-paper version, on-line version, other added-values), which should be sold as options, rather than used to try to keep holding the essentials (the refereed final draft) hostage to access-tolls.

There will still be a permanent niche for journal publishers. What remains to be seen is whether that will entail downsizing to peer-review service-provision alone, or whether there will also continue to be a market for toll-based add-ons even after the refereed drafts are available free through the Eprint Archives.

7.6. Government/Society: Mandate public archiving of public research worldwide

The beneficiaries will not just be research and researchers, but society itself, inasmuch as research is supported because of its potential benefits to society. Researchers in developing countries and at the less affluent universities and research institutions of developed countries will benefit even more from barrier-free access to the research literature than will the better-off institutions, but it is instructive to remind ourselves that even the most affluent institutional libraries cannot afford most of the refereed journals! None have access to more than a small subset of the entire annual corpus (http://fisher.lib.virginia.edu/arl/index.html). So free access to it all will benefit all institutions (Odlyzko 1999a, 1999b).

And on the other side of barrier-free access to the work of others, all researchers, even the most affluent, will benefit from the barrier-free impact of their own work on the work of others. Moreover, a freed, interoperable, digital research literature will not only radically enhance access, navigation (e.g., citation-linking) and impact, hence research productivity and quality, but it will also spawn new ways of monitoring and measuring that impact, productivity and quality (e.g., download impact, links, immediacy, comments, and the higher-order dynamics of a citation-linked corpus that can be analyzed from preprint to post-postprint, to yield an "embryology of knowledge" (Harnad & Carr 2000).

To put this worry into perspective, we must remember that print-on-paper is not permanent either. The only relevant parameter is the probability of future access. The on-paper probability, such as it is, is achieved by generating (a) multiple copies, (b) geographically distributed,  (c) in a (relatively) robust medium, (d) visible to the human eye.

All four of these properties can be (and have been) achieved on-line too, and the resulting preservation probability can be made as good as, or even better than, the current probability on-paper.

That should be the end of the story: For once this concern is no longer grounded in actual, objective probabilities, but only in prior habits and attendant intuitions, then we are talking about biasses and superstitions and not about actual risks.

There are a few side issues: People worry about global power-failures, or global dictatorships. They should remind themselves that these are matters of probability too, and have their equivalents in paper.

People also, by analogy with current unreadable documents in obsolete word-processors or peripherals, worry about whether the digital code, even if preserved, will always be accessible and visible to the eye.

The answer is again probability: The reason print-on-paper has been faithfully preserved across generations (when it has been) is that the literate world's collective interests were vested in ensuring that it should do so. This same continuity of collective interests will exist for the digital corpus too, for the same reasons, except that digital code will be much easier to keep uploading to every successive new technology than print on-paper to every successive building or regime ever was.

(And there is always the option for those who are still not confident enough in the technology, despite the odds, of printing out hard copies as back-up: Indeed, that is a good way to put the magnitude of one's Zeno's worries to the test: Who will still feel the need to make hard copies, and of how much of the corpus, once it's all on-line and accessible to everyone, everywhere, at all times?)

In short, preservation measures as a practical pursuit by digital librarians is an eminently worthy one; but as a basis for any hesitation or delay whatsoever about proceeding with self-archiving right now, it is completely irrational (particularly as, for the time being, self-archiving is merely a supplement to, not a substitute for, the existing Gutenberg modes of preservation).
 

Again, the rational way to put this into context and proportion is to remind ourselves that the authenticity of an on-paper version is just a matter of probability too, and that the very same factors that  maximize that probability on-paper can maximize it on-line too. Indeed, if we wish, we can make both the probability and the verifiability of authenticity on-line much higher than it currently is on-paper through techniques such as public hash/time-stamping and encryption.

Nor should the authentication issue be confused with the issue of Peer-Review (7) or Journal Certification (5) (separate questions), nor with the question of "version control" (there will be self-archived preprints, revised drafts, final accepted, published drafts (postprints), updated, corrected post-postprints, peer comments, author replies, revised second editions. In all of this, the refereed, accepted final draft is one crucial "milestone," but not the only one, in the embryology of knowledge (and not even always the best one).

And last, some of the "authentication" worries arise from conflating self-archiving and self-publication. To say it in longhand: The main objective of the self-archiving initiative is the freeing of the refereed drafts from access/impact barriers. The refereed draft has already been "authenticated" by the journal that peer-reviewed it. Do not confuse that authentication with some worry you may have about whether this self-archived draft is indeed what the author purports it to be. The only thing the author is "self-certifying" in this case is that this is indeed the journal-certified final draft. There is of course always a possibility that it is not the journal-certified final draft; but that was also true when the author sent you an on-paper reprint. The probabilities can, as usual, be tightened to make them as high as we feel comfortable with in either case. And in the case of preservation, self-archiving is at this stage merely a supplement, not a substitute for existing forms of authentication.

So, again, there are no rational authentication concerns whatsoever to deter us from self-archiving immediately.
 

Again, the answer is that simple and effective means are available to ensure that an on-line draft is uncorrupted with as high a probability as we feel we need. So this too is a non-problem. (Nor should it, again, be conflated with self-publication issues, which are irrelevant to the self-archiving of refereed, journal-published papers.) Whatever level of incorruptibility we feel we need, we can have it for self-archived papers too.

Consequently, corruptibility worries provide no rational basis whatsoever for deterring us from self-archiving immediately.

The primary objective of self-archiving is to free the refereed journal literature from access-tolls on-line. That literature is already being published  on-paper. (If you think it should not be, it is with the journals and their referees that you need to take issue, not with self-archiving or the on-line medium!) When it is all accessible free on-line, there is no need for anyone to feel any more (or less) obliged to read the refereed literature than they did on-paper. Keeping it off-line is certainly no cure for the information glut (if there is one); it merely makes the existing access-tolls the arbitrary arbiters of whether or not one reads something, rather than the reader's own rational judgement. (And unrefereed preprints can of course always be ignored altogether, if the reader wishes, on-line just as on-paper.)

In short, no rational deterrent at all to immediate self-archiving from concerns about navigation or information glut.

Again, no rational deterrent to immediate self-archiving in the certification worry.

So there is no rational deterrent to immediate self-archiving anywhere in the evaluation worry.

No rational deterrent to immediate self-archiving in the peer-review worry.
 

There are many fallacies embedded in this worry, among them misunderstandings about the nature of global networked communication. Internet connectivity, at very low cost, is now part of the infrastructure of most of the world's universities and research institutions. If you are not equally worried about who pays for your emails, websites, and web-browsing, you should not be worrying about your self-archiving either. In any case, paying access-tolls is not paying the pertinent piper here anyway!

The refereed research literature is minuscule compared to the rest of the traffic on the Web (http://www.sims.berkeley.edu/how-much-info/summary.html).It is the flea on the tail of the dog. Worry about the storage and band-width for the growing daily creation and use of audio, video, and multimedia (most of it non-research use!) by researchers at universities and research institutions before even beginning to fret about the refereed flea. As usual, there is also some of the archiving/publishing conflation here, thinking that we must find some sort of counterpart for the printing/distribution costs, somewhere. But there isn't any. The price per-paper of permanent online archiving is virtually zero, yet everyone, everywhere, has access to it all, forever. This is a Gutenberg expense that has simply vanished in the PostGutenberg Galaxy, leaving only the Cheshire Cat's Grin.

There is indeed one essential publishing cost that still needs to be paid, but it has nothing to do with Internet use: It is the cost of implementing peer review. That cost, however,as discussed in the Peer Review section (3.3), is only 10-30% of the access-toll costs currently being paid, and hence could easily be paid out of the annual savings.

The last of the "who-pays-the-piper" worries is, I think, a variant of the Capitalism (14) worry. The best way to dispel it is is to note that refereed publishing in the PostGutenberg Galaxy, once the literature has been freed through self-archiving, is likely (apart from whatever optional add-on products and services there may still be a market for) to downsize into a service (peer review), provided to the author-institution, instead of the toll-based product (the text) that was provided to the reader-institution in the Gutenberg era.

Nothing hinges on this, however, for as long as the world wants to keep paying for the toll-based product, even after the refereed literature has been self-archived, the piper will be fully paid, yet the literature will be free of all its access/impact barriers.

No rational deterrent to immediate self-archiving in the who-pays-the-paper worries.
 

No one can predict with certainty the evolutionary path that scientific/scholarly journal publishing will take once the refereed corpus has been freed online by self-archiving. The toll-based market for the on-paper version, for the publisher's on-line version or for other options may continue indefinitely, or it might shrink but re-stabilize at a lower level, or it might disappear altogether -- and this could happen relatively slowly or relatively quickly.

It is not clear in advance which of the current established journal publishers will want to continue doing what, under what conditions. The bottom line is that the only remaining essential service will be peer review. If and when that is the only service for which there remains a market, either current journal publishers will be able and willing to downsize to that niche, or they will terminate journal operations, in which case their titles (that is, each journal's editor, editorial board, referees, and authorship) will simply migrate to new on-line only journal publishers who are ready to adapt to the new niche [e.g., the Institute of Physics's New Journal of Physics (http://www.njp.org/) and BioMed Central (http://www.biomedcentral.com//)].

No rational deterrent to immediate self-archiving in worries about publisher downsizing.
 

In brief, many journals will agree to author self-archiving if the author asks, and for those that don't, self-archiving the preprint before submission and a "corrigenda" file after acceptance is sufficient, and completely legal. What career and livelihood depend on is peer review and impact, and all self-archiving authors continue to have both; neither needs to be sacrificed for the other.

No rational deterrent to immediate self-archiving in copyright worries.

Depending on how important we find it to do so, we can make escape from detection so improbable on-line that it becomes harder to plagiarize on-line than on-paper. It is not clear, however, whether it is even all that important to do so. Worries about plagiarism are usual based on the archiving/publishing conflation: Once one's findings have been refereed and published, it is hard for anyone else to derive any benefit from them at the expense of the author (the peer-reviewed version settles all subsequent authorship disputes).

Pre-refereeing preprints are another story; they are dealt with partly in the prior discussion of Authentication (2), and partly under Priority (12), below.

For refereed postprints, however, refraining from self-archiving them because  of worries about plagiarism would be no more rational than refraining from publishing them on-paper in the first place, for the very same reason.

No rational deterrent to immediate self-archiving in priority worries.

It is true that one's on-line literary goods are at the mercy of the archives and archivists. But one's analog on-paper literary goods were likewise at the mercy of the libraries. They could have chosen to "censor" our work too.

Again, it is just a matter of deciding how tight we wish to make the probabilities in this medium. Mirroring, caching/harvesting and distributed coding already go some way toward taking it out of any potentially sinister local hands.

No rational deterrent to immediate self-archiving in worries about censorship.

Nor is there any market interference in self-archiving one's own refereed research: If institutions and individuals want to pay for toll-access to the on-paper version, or the publisher's PDF, or further options, they can still do so; but there is no longer any need or justification for continuing to hold the essentials (the peer-reviewed draft) hostage to those toll-based options in the PostGutenberg era, any more than there was any need or justification for continuing to hold the essentials of long-distance communication hostage to postal transport costs in the era of telephony. (Rather than capitalism being under assault from self-archiving, trying to prevent researchers from benefiting from this new, more efficient and economical way of disseminating and maximizing the impact of their refereed research smacks of protectionism.)

Two variants on the capitalism-worry arise from scepticism about the eventual transition from providing an toll-based product to the reader-institution to providing a peer-review service to the author-institution. Note that, strictly speaking, it is not even necessary to answer these worries, as this eventual transition is hypothetical, whereas freeing the refereed literature now through self-archiving is not; but here are replies anyway:

Question 1: "Won't paying directly for the peer-review service lead to inflated peer-review costs by the most prestigious journals?"

Question 2: "Won't peer-review revenues lower standards, so that lower-quality work is accepted in order to get more peer-review revenue?"

The answer to both is similar: Referees referee for free, and journal quality and prestige (and impact) depend on rejection rates. Trying to inflate revenue by lowering acceptance thresholds simply lowers quality, thereby favoring the competition, with higher standards. It is a built in counter-weight. Likewise for raising peer-review rates: As referees referee for free, there is no reason one journal should charge more than another, and if they do, they risk driving not only the authors but the the unpaid referees to the competition. Because the competitive commodity in this anomalous give-away domain is quality, and nothing else.

A proposal has occasionally been voiced to keep preserve toll-barriers by buying authors off from self-archiving, by offering to share the revenue with them (royalty payments). But the trade-off between imprint-income and impact-income is so disproportionate for this anomalous domain that there is not faintly enough money available to make others prefer sacrificing their potential impact in exchange.

No rational deterrent to immediate self-archiving in worries about capitalism.

No rational deterrent to immediate self-archiving in worries about readability.

No rational deterrent to immediate self-archiving in worries about graphics.

No rational deterrent to immediate self-archiving in worries about publishers' future.

The serials literature is all going on-line anyway, irrespective of the speed or success of the self-archiving initiative. If this requires restructuring of some librarian skills and functions, this will take place in any case. Some have thought that managing digital serials collections will fill the gap, but it is not clear how much management those will need, apart from paying the annual toll-bills! Author/Institution Eprint Archives, on the other hand, will call for more digital librarian skills, in everything from helping researchers to do the self-archiving, to maintaining the institution's Eprint Archive and seeing to its continued interoperability with the rest of the world's Eprint Archives, its upgrading, and its preservation.

Moreover, in implementing and maintaining the institutional Eprint Archives, Libraries will be investing in the solution of their serials crisis. Of the 100% annual toll budget that this can potentially save, after 10-30% of it has been redirected to cover author-institution peer-review costs, the remaining 70-90% can be used to fund other librarians' activities, including the purchase of non-give-away materials such as books (whether on-paper or on-line).

No rational deterrent to immediate self-archiving in worries about libraries'/librarians future.

But many of them are also journal publishers, and hence may be facing downsizing pains. Unlike commercial publishers, however, their first and last allegiance will of course be to research and researchers, that is, us. We will hear rationalizations about needing the toll revenues to fund "good works" such as meetings, scholarships and lobbying. But it will quickly become evident that, on the one hand, some of these good works are not essentials either, and certainly nothing that we would want to sacrifice research impact for; and the subset of them that really is essential (such as meetings) will prove to be able to fund itself other ways too, rather than needing to be subsidized at the expense of research impact.

Learned Societies (and perhaps also University Presses) are also natural candidates for taking over the serials titles of commercial journal publishers who prefer to discontinue journal operations rather than scale down to just becoming peer-review service providers.

No rational deterrent to immediate self-archiving in worries about Learned Societies' future.

We should not forget that the give-away refereed literature is esoteric, with virtually no "market" per paper. So whereas there might be a basis for suspicion about what our hard-pressed universities might like to do if they could get their hands on our exoteric, non-give-away work (royalty-bearing books and textbooks), there's not much they could do to squeeze revenue out of our no-market, give-away refereed research reports even if they wanted to. On the contrary, our universities, like ourselves, benefit far more from the potential impact-income of such work -- maximized by removing all access-barriers -- than from any potential imprint-income that could be squeezed out of it by co-opting the "P" from the publishers' S/L/P tolls and using it to charge institutional archive access-tolls.

Moreover, our universities' potential toll savings, and relief from their serials crises, are completely dependent on freeing access to our research. Any sign of university-levied archive-access tolls would simply serve to keep the current access-tolls in place.

No rational deterrent to immediate self-archiving in worries about University conspiracy.

No rational deterrent to immediate self-archiving in worries about loss of serendipity.

The other half of this worry is probably a variant of the Copyright (10) concerns (q.v.) as well as concerns about Embargo policies (Harnad 2000a, 2000b), both of which are groundless.

No rational deterrent to immediate self-archiving in worries about tenure/promotion.

9.1 Napster

It is very important to clearly distinguish and distance the two, because any inadvertent or willful conflation of the self-archiving initiative with napster can only retard the progress of the self-archiving initiative toward the optimal and inevitable.

("Information is free" is nonsense: There is and always was both give-away and non-give-away information. Steal the latter and you simply kill the incentive to provide it in the first place.)

9.2 Peer-review reform

Hence current peer review reform or elimination proposals are merely speculative hypotheses at this time, and red herrings insofar as the freeing of the peer-reviewed literature is concerned: The self-archiving initiative is directed at freeing the current peer-reviewed literature, such as it is, from the impact/access barriers of S/L/P access-tolls, now. It is not directed at freeing the literature from peer review, or at testing or implementing untested alternatives to peer review (Cf.
http://library.caltech.edu/publications/ScholarsForum/042399sharnad.htm
and http://www.cogsci.soton.ac.uk/~harnad/Ebiomed/com0509.htm#harn45).

The benefits of freeing the refereed literature now are a sure thing; the benefits (if any) from future alternatives to peer review (if any) are purely hypothetical, and certainly nothing to hold as back from self-archiving to wait for.

9.3 "Scholarly Skywriting"

9.4. Embryology of Knowledge

9.5 Leading horses to the waters of self-archiving vs. getting them to drink

10. APPENDIX A: Eprints.org Software for Creating Institutional and Individual Open Archives

At Southampton, we took this to heart, and applying our experience with the CogPrints archive, designed the generic eprints.org software that fits this bill. A public beta version has been released and has taken over operations at the CogPrints site <http://cogprints.soton.ac.uk/>. The operational release (December 2000) is free and will be open sourced (and over 100 prospective users worldwide are already signed up).

The eprints.org software is a feature-rich, easily installed, eprint archive system. It runs right "out of the box" with a comprehensive default setup that should serve most institutions' and individuals' needs as it stands. It has also been designed to make it extensively and flexibly re-configurable for customised needs; almost any aspect of the archive's operation can be adapted to suit a particular requirement.

The archive supports the OAI protocol, allowing it to interoperate with other open archives and open archive services, and to be readily upgraded to keep up with OAI revisions.

This adaptability is achieved by using a modular design methodology. The system is divided into two main components: The core archive component, which provides the functionality required for all open archives, and the site-specific component, providing details about exactly what is stored in the archive, how it is presented and how it may be searched. The system is supplied with a richly featured site- specific component that requires minimal changing to set up a fully working, interoperable open archive. When updated revisions of the software become available, the core archive component can be upgraded, and the site retains its identity and data in the site-specific component.

    10.1 The many aspects of the software that can be configured by an institution include:
 

•         The types of record that can be stored in the archive, and what metadata fields to hold with each

•         The types of document file (or other data) that can be stored with each record

•         The validation checks that are performed on each incoming record, to minimise administrator effort

•         The choice of which metadata fields are searchable by users

•         The choice of what metadata to present records to the open archives protocol (i.e., specifying how the internal metadata maps to the open archives metadata)

•         Full control over the "look and feel" of the archive (in any language: French translations currently being prepared by Helene Bosc and colleagues at http://inra.fr )

       10.2  The software also has the following features:
 

•         "Out of the box" Open Archives Initiative interoperability

•         Simple but very powerful depositing interface

•         Local browsing and searching features

•         Inter- and intra-linking potential (papers, versions, comments, responses)

•         Moderation buffer for incoming deposits

•         Site maintenance via a WWW interface

•         E-Mail subscription service for users

    It is simple to add extra functionality to an archive in the site-specific component of the software. This means that the archive can be used by institutions, individuals, journals or any other organisation wishing to interoperate with Open Archive services.