Comments on E-Biomed -- 07/12

> That the desire for acceptance in the full-blooded sense is the
> motivating factor underlying quality in preprint publication, is a
> testable hypothesis, I would think, though some may think it too
> obvious to require testing.

On the contrary: As I said in a prior response, this "hand-on-heart" substitute for classical quality control would decidedly need to be tested to be taken seriously. The road to Usenet is paved with good motivations...

> ... the reason why
> it works for those fields is that the people who are there to respond
> to a preprint publication via the server are peers who are working at
> the leading edge of the field and who recognize that what is made
> available there via the server is to be treated as primary publication
> and responded to accordingly.

I will await more systematic and empirical analyses of the remarkable historical developments in physics and preprint self-archiving in this decade before drawing conclusions about why it happened in physics first, what may or may not be unique to physics in it, and what relation it may have to the refereed physics literature, or the need for it.

ONE empirical hypothesis seems worth testing already: Will the value of PREprint self-archiving generalize to the rest of the disciplines?

And one obvious LOGICAL implication of the fact of LANL does not even require testing, but calls for immediate APPLICATION now: The value of (refereed) REprint self-archiving -- freeing the refereed literature online -- WILL generalize to the rest of the disciplines.

> Paul Ginsparg did not create the community of
> preprint users that is the underlying reality of the science the server
> system at LANL serves...
> Thus neither NIH nor the BMJ nor the Caltech people have it within their
> power to duplicate the Ginsparg achievement: it can't be duplicated by
> building archives and inviting people to use them, but only by finding
> existing practices, if there are any, that can benefit from the use of a
> system like this and enabling them to do so.
> ... if a field is not coherent and mature enough in its pre-existing
> practices to use a preprint server effectively as a means of primary
> publication, it just will not work...

Too many inferences and interpretations here. Here is what can be said with confidence: LANL has shown that PREprint self-archiving would be a promising thing to try in other disciplines too, and REprint self-archiving would be a sure thing to succeed in other disciplines too.

> Much of your response assumes that you have to defend peer review. But
> I haven't questioned its importance or validity at all so there is no
> reason for me to respond to that part of it.

Implicit in your interpretations about what is going on in LANL, and where it is going, are assumptions (in my view, incorrect ones) about the causal role of peer review in all this. I have simply made those assumptions explicit, along with the evidence and arguments against them.

Stevan Harnad
Professor of Cognitive Science
Department of Electronics and Computer Science
University of Southampton
UNITED KINGDOM
http://www.cogsci.soton.ac.uk/~harnad/
http://www.princeton.edu/~harnad/
ftp://ftp.princeton.edu/pub/harnad/

Mike Mage, July 15, 1999

Dear Dr. Varmus,

I may have missed it, but your proposal does not seem to recognize that in a way, E-Biomed is here already. The number of personal websites is expanding exponentially, and there are no obstacles now to scientists posting whatever they want about their research on personal websites. Perhaps E-Biomed could best serve as an index, electronic discussion board, and search engine for totally decentralized posting of research results. Thus, If I posted research results on my personal website, E-Biomed could be a designated place for anyone else to post comments or criticisms of the work. This would be a democratic and fair way to be helpful that would not step on any toes.

Sincerely,
Mike Mage

July 14, 1999

Stevan Harnad, University of Southampton, July 14, 1999

On Tue, 13 Jul 1999, ransdell, joseph m. wrote:

> There is surely no question but that the significance of LANL is
> commonly thought to lie in its success as a preprint server, not in the
> many other facilities added or being added, and it is reasonable for
> people to think that in adopting it as a model you are doing so because
> of that for which it is famed, not for other features of it of no
> special interest as innovations.

It is a historical fact that LANL begin as a preprint distribution network among 100 high energy physicists. It is a further historic fact that it rapidly grew to encompass more and more authors and users, covered more and more of physics and beyond, and came to include the refereed final drafts too. If we are going to take LANL as a model -- as we should -- let's make sure we use the full current model and not just the barest initial conditions!

And the fact is that the "other feature" (entirely predictable if the rapid growth of the scope and scale itself had been predictable) -- namely, that it ain't just about preprints any more, but is rapidly growing into the WHOLE PHYSICS JOURNAL LITERATURE -- is turning out to be (in my view) even more important than LANL's preprint function (though that continues to be very important too), because it demonstrates the revolutionary possibility of freeing the journal literature for once and for all, across disciplines and around the world, to the eternal benefit of Learned Inquiry through SELF-ARCHIVING.

So self-archiving is the "model" and the take-home message of LANL, and not merely, or primarily, the self-archiving of unrefereed preprints.

> The appeal to the invisible > hand does not lessen the import of the fact that the LANL system uses
> unfiltered material, and it is bound to occur to many people that there
> is not, after all, any rule that requires people to simultaneously
> submit the paper to a referee: that is just a custom at LANL at best.

Well, isn't it odd that today, when LANL is up to 20,000 new papers annually, that that alternative still does not seem to have caught on, and Physics journal submissions continue to proceed apace?

Nor is it difficult to see why: For vanity publication (I am afraid I must persist in the pejorative usage, particularly in this context, to stress precisely what is at issue) does not count very much toward the promotion of either one's findings or one's career. In general, the academic reward system relies on reliable, credible quality control and certification -- as do the users of the literature (which, as I keep recalling, is still both constrained and sign-posted by the invisible and visible hand of peer review); it is not only promotion committees, but readers who would be overwhelmed and helpless without it.

The preprint dimension is a splendid, indeed revolutionary supplement to the classical system: The scope and scale on which new findings and ideas can be disseminated immediately, even before being vetted for their quality and reliability by experts, is exhilarating and will no doubt increase the scope and scale of Learned Inquiry -- although I don't think it will do see nearly to the degree to which the even more revolutionary feature, the freeing of the refereed corpus, will. And the interactive possibilities -- commentary and peer commentary, on work both before and after peer review -- will be revolutionary too.

But quality control must persist -- and along the classical lines (until an alternative that does at least as good a job is first found and tested): The invisible hand of peer review must continue to be made visible, reliably sign-posting the literature for the otherwise hapless Hitch-Hiker in the PostGutenberg Galaxy, be he a member of a promotion committee or just a journeyman researcher trying to contend with the swelling literature.

> The reason I am pushing this to the fore is that I notice that you don't
> any longer seem to regard the preprint server as an important part of
> it. When the immunologists responded in tones of outrage to precisely
> that feature of the E-biomed model you quickly advised the NIH people
> not to worry about implementing the preprint server: that could be
> figured out later; what is important is the refereed literature,
> anyway.

I do consider the self-archiving of preprints to be extremely important and desirable. I was responding there to the special case of CLINICAL MEDICINE (in the context of the E-biomed initiative), where public health might be at risk from wide distribution of unrefereed claims.

I actually believe that the clinical community will be able to set up reliable safeguards without too much difficulty -- public-health vetting that is short of peer review but filters out dangerous errors and quackery from the unrefereed preprint sector of the clinical portions of the biomedical literature in E-biomed. And the "R" (for refereed; and "JX" for journal-name) sign-posts will also help to distinguish what is safe to take seriously in the clinical literature. (I even suggested, tongue-in-cheek, that a cigarette-like "potential health hazard" tag could accompany all unrefereed clinical preprints too).

The option to "forget about preprints for now" was hence only directed at clinical researchers, and only to dispel the red herring of public health risk as a rationale for opposing E-biomed self-archiving simpliciter: For the REFEREED clinical literature is certainly no health hazard (one hopes!), and hence should be self-archived forthwith...

I do not believe that the special case of this (soluble) health-risk problem for the clinical literature generalizes to the literature as a whole.

> > ... it [the invisible hand] constrains preprints to be
> > drafted on the presumption of answerability to classical
> > peer review, through conventional journal submission,
> > usually concurrent with archiving.
>
> What constraint? What presumption? The constraint on the author is
> that what he or she writes is to be in agreement with the facts, as
> these are ascertained in the course of inquiry, not as they are
> established through agreement with peer reviewers. The invisible hand
> is just an awareness of a future contingency that can be handled in more
> than one way or even ignored...

How much easier my job as Editor would be if all that authors had to do was hold their hands to their hearts and state that what they have written is sound, in conformity with the facts, competent -- in short, worthy of the finite reading time of every busy researcher attracted by the title who, in the old days, would have had that limited reading time guided by the intrepid experts who had first done the hard interactive work of getting the promising papers ship-shape and certifying them as such!

> I must say that the resort to the concept of
> the "vanity press" seems to me just gratuitously contemptuous of people
> and their motives and is very misleading as regards what is actually
> happening when people try to communicate. How can we implement a
> communications revolution with the use of simplistic denigrations like
> this?

You know what they say about good intentions! But to change the metaphor: would you like to choose your daily food (or drugs) not after prior FDA vetting, but... what? post-hoc opinion polls?

Richard Hill, Ph.D., Geneva, Switzerland, July 14, 1999

This is a fabulous idea, long overdue. I really hope that you will go ahead and do it.

Richard Hill, Ph.D.
Geneva, Switzerland

July 13, 1999

Stevan Harnad, University of Southampton, July 13, 1999

On Tue, 13 Jul 1999, Bob Parks wrote:

> 1. xxx.lanl.gov has about 100,000 papers and that archive does not seem
> to have reduced the number of journals in physics, nor the quality of
> the scientific literature. Hence we have at least one strong piece of
> evidence that 'free access archiving' will not lower the quality. I
> don't know of any evidence showing that quality has been lowered in
> physics or elsewhere.
>
> 3. xxx.lanl.gov seems to have conditioned its audience to 'filter'
> relevent articles from the large number of submissions. I would guess
> that works much like the usual filtering process that any academics use
> for 'hard copy working papers'.

The filter is even simpler than that: "R" (refereed journal, and perhaps journal name "JX") and "Author Name."

The peer review proficiency of the journals in question takes care of the rest.

> 4. When we have citation-linking for all scientific literature
> http://www.princeton.edu/~harnad/citation.html it will be natural
> and easy to 'value' writing - namely by the number of citations (and
> possibly the 'quality' of citations). Such citation criteria are
> already used in promotions and salary (at least in my small biased
> sample). One can argue whether quality is better determined from
> citations than from knowing that two or three referees and an associate
> editor have passed judgement.

Fallacy: Apart from the limitations of citation metrics, the validity they do have is COMPLETELY parasitic on the fact that the papers in question are published in peer reviewed journals. The invisible hand of peer review is behind them. Hence they would implode if that hand were withdrawn.

Impact factors (like peer commentary) are a SUPPLEMENT TO, not a SUBSTITUTE FOR, peer review. (And peer review is not a go/no-go filter; it is an interactive, iterative, corrective process of submission, feedback, revision, resubmission, etc. between author and peer reviewers, adjudicated by a competent, answerable peer Editor: This is no passive numerical filter or box score on any blind metric.)

Harnad, S. (1998) The invisible hand of peer review. Nature [online] (5 Nov. 1998) http://helix.nature.com/webmatters/invisible/invisible.html
http://www.princeton.edu/~harnad/nature2.html

> 5. As an economist, I would have to argue that the resources devoted
> to refereeing are misallocated because they are not compensated
> directly. In the current journal model, there may be too much
> refereeing (or there may be too little).

Peer review, like democracy, is not without its imperfections. And no doubt there exist ways to improve it. But those ways must first be found and tested and demonstrated to improve it -- especially the most radical one, of abandoning it entirely, in favour of self-archiving of unrefereed preprints alone.

Note that I do NOT advocate the latter: I have always aimed (subversively) at the self-archiving of the REFEREED paper, not just the unrefereed one. That will be tantamount to freeing the current peer-reviewed literature, SUCH AS IT IS (warts and all).

http://www.arl.org/scomm/subversive/toc.html

We can worry about ways of fixing the warts independently; but the powerful and proven benefits of self-archiving should NOT be linked in any way to speculative and untested notions for improving peer review -- least of all abandoning it altogether.

> If 'free access archiving'
> means the end of journal refereeing as we know it, I am not sure
> whether I (at least) could argue that there is a social gain or loss.

There is every reason to believe it would be an enormous loss, throwing the baby (a reliable research literature) out with the bathwater (access barriers to that very literature).

To put it another way, it's READER-ACCESS barriers that have to go; AUTHOR-ACCESS barriers (into the certified refereed corpus) must stay.

> Referees might spend their time writing/reading rather than refereeing,
> which could result in better scientific literature than what exists
> with their time spent refereeing. I am not arguing that refereeing has
> no value, only that we do not know what that value is, and that
> whatever that value is, it is not compensated (directly at least).

We know the value of the refereed literature, such as it is; every editor knows what raw submitted manuscripts look like (90% of which will be rejected, if it is a high quality journal; and most of the 10% that are finally accepted will look nothing like their initial drafts, for they will have gone through the iterative corrective feedback cycle of peer review mentioned above).

It is the difference between these two literatures that is at issue (and the difference is even greater than that, because even those raw submissions are prepared with the PRESUMPTION of answerability to peer review -- yet another manifestation of its "invisible hand").

No, human nature being what it is, without answerability it quickly regresses toward the anarchic levels of the chat-groups NetNews, that Global Graffiti Board for Trivial Pursuit. (And neither the Pandemonium of post hoc "peer" commentary nor the still poster-hoc feedback from "citations" can provide that answerability. Pity the reader who has to navigate a chaotic corpus like that.)

> IMHO, the only reason to sort it out is to determine, given the goals
> of the esoteric author (a term I like), whether 'free access archiving'
> will lower or raise the quality of scientific literature.

IMHO, it would not be good empirical practise to test whether the FDA is really protecting our healths by scrapping it and seeing what happens!

The goal of self-archiving is to free the refereed literature from access barriers, not to free it from refereeing!

The way to test variants on or alternatives to peer review is locally, not globally: Do you know of any local experiments? I do, and as far as I know, it is not yet faring too well!

http://www.bmj.com/cgi/shtml/misc/peer/index.shtml

> Again, the point should be whether the quality of the scientific
> literature is harmed by 'free access archiving'....
> In the NO-JOURNALS world of 'free access archiving'
> we write to attract others attention, and citation. Rather than
> writing for three people (two referees and an editor) we now have to
> write for a larger audience and have to write to attract a readership
> (rather than attract an editor/referee). I don't see that deters us
> writing. The goals of fortune and fame remain, its just the journals
> no longer have a Faustian GRIP on us.

And it all becomes a vanity press, with no sign-posts for the poor reader and user as to what, in all this unregulated soup, is fit for consumption!

> The current business 'model' for scientific literature is, well,
> absurd. Editors are mostly not directly compensated, and those who are
> are not compensated at the market value of their time. Referees are
> not compensated ($35 or $50 is not compensation). Authors are not
> compensated at all directly.

Correct. But it is also what vouchsafes us our current refereed literature, such as it is. Let us free THAT before toying with any notional improvements -- INCLUDING referee payment, which is potentially corruptive: They referee for free now, and that's just part of the system, such as it is.

The system, with its authors giving their papers away free, and its referees giving away their services for free, is best described as ANOMALOUS, not ABSURD. What is absurd is to continue treating it according to the access-blocking trade-model, instead of freeing it (from S/L/P) for the reader as well.

Here is an argument that could be invoked against me, but I don't think it's valid:

"Fine, I take you at your word. 'Don't tamper with the system, just free it.' Now I will show that that admonition is self-contradictory: I agree that altering or abandoning refereeing would be tampering, and would put quality at risk. I agree that paying referees would be tampering, and would put quality at risk. But then, isn't author self-archiving tampering too, and putting quality at risk? Might it not bring down the entire system, destroying the revenue base on which the quality control is built? Is it, therefore, not one of those untested "reforms" of peer review against which you always inveigh? Q.E.D."

My reply is simple: Authors have always given away reprints of their papers for free. Self-archiving simply increases the scale of this. And the waters HAVE been tested, for close to a decade now, by Los Alamos, and no sign of diminished quality has emerged (as Bob notes above).

http://xxx.lanl.gov/cgi-bin/show_monthly_submissions

So the free waters are safe for peer review. (And if and when another revenue source must be found for continuing to fund it, the up-front redirection of 1/3 of S/L/P savings to institutional publication costs is ready as a natural source for it.)

Hal Varian's worries about possible knock-on effects elsewhere in the quality control system are not entirely without basis, but there doesn't seem any compelling reason for alarm either. Let self-archiving proceed apace.

> So the university pays us to
> author/edit/referee and then buys our product back from a 'publisher'.
> Resources must be misallocated in that model. If our current world was
> a 'free access archiving' with citation valuations (rather than journal
> valuations), proposing such a business model would, well, be absurd.
> We need to unshackle ourselves from the current journal Faustian Grip,
> from that mental model of the world, and proceed ahead. Nor should we
> consider that scientific literature fits into other 'information'
> products.

This seems a bit garbled to me, because it conflates freeing the refereed literature form S/L/P with freeing the literature from refereeing.

> Much of the discussion between Hal and Stevan side steps into business
> models (ignoring any further words on motivations of authors). So what
> is the business that requires a model? Production of (quality)
> scientific literature.

Actually, the "business" is making an impact on research with one's research. The literature is just a means, not an end. Hence my analogy with ads.

> Must that be tied to the elsevier et al (I use
> elsevier in lower case as a generic for profit and non-profit
> presses)? elsevier does not pay the authors, nor the referees nor the
> editors which is 95% to 99.9% of the real cost of producing the
> literature. In the 'free access archiving' world, we do not need to
> wory about whether elsevier survives. We do need to worry about the
> quality of the scientific literature, and elsevier itself does not
> provide that quality control. Editors and referees do. Citations do.

Peer review does; citations do not! Elsevier (and others) implement peer review. That will always cost something -- but nothing like what the whole papyrocentric, S/L/P-based system costs now.

> Whether universities are willing to compensate us for editing and
> refereeing without the elsevier label is an open question (especially
> if the citation linking proposal becomes fact). In fact it is a
> question which should be asked - how much refereeing should be done?
> If we have citations, do we need refereeing and editing? It is not
> that refereeing and editing do not increase the value of an article, it
> is whether the correct amount of resources are devoted to that
> activity, and whether citations (or similar) would be a more cost
> effective way to discern the quality (for promotion, tenure, etc.).

Vide supra. It seems to me naive in the extreme to imagine that delayed post-hoc citations can substitute for the substantive, interactive, quality-control process of peer review.

> Imagine a world with 'free access archiving' without journals. How
> does one get promoted? Citations and review letters. Citation
> analysis would be free, and universities would have to compensate for
> outside review letters. Would that really change the quality of
> scientific literature - for the worse? Not in my mind.

Have a peek at Usenet/Netnews for a glimpse of where things would head, human nature being what it is...

David M. Maurice, Ph.D., Columbia University, July 13, 1999

Dear Dr. Varmus,

I would like to congratulate you on your e-biomed initiative and I would like to suggest some extensions that would make it more democratic and require less central control.

First, I would like to point out that, nowadays, open discussion of scientific issues is virtually non-existent, at least in the fields with which I am familiar. Many publications I read seem to contain imperfections either minor, e.g. a failure to cite some evidence contrary or in favor of the author's conclusions, or major e.g. a flaw in their experimental techniques that would vitiate their findings. However, critical letters to the Editor are very rare; writing a letter has become a big deal, and anyone who publishes one as much as once a year is likely to be regarded as a crank. The possibility of open discussion at meetings has also diminished now that posters are becoming the dominant form of communication.

With e-biomed, it would be possible for a more relaxed publication of queries, criticisms, or even compliments, to become routine and productive discussions could arise, not subject to arbitrary termination by the journal editors. I do not think this would be subject to abuse; intelligent, superficial, or rancorous, correspondents could be distinguished by the reader and the views expressed judged accordingly. Furthermore, the option is available of initially corresponding privately with the author leaving it open for the critic to voice his opinions openly if he is not satisfied; on other occasions the parties may reach a resolution which they could publish jointly if they thought it sufficiently important. Unlike anonymous refereeing, open criticism would have to be fully responsible and the authors would have an opportunity to reply.

Another benefit of a more casual attitude towards commentaries on publications is that it would afford some relief to those who feel that their own work is being unjustly neglected. When a relevant article is published they could present appropriate sections from their own publication. If they persevere to the point of seeming foolish, so be it.

Second, I have the impression that there has been a marked drop in the quality of presentations of the manuscripts that I have handled in my 50 years as a reviewer. Some authors seem to throw their results together almost anyhow, just to get a preliminary response from the editors, I suspect. This results in a lot of extra effort from a conscientious reviewer, and I find I am spending more time in rewriting the papers of others than writing my own. This burden would be lightened if unedited publication was the rule, and then the community would be able to judge the quality of the author's work in the raw.

In addition to signed (and possibly unsigned) commentaries, a scheme can be envisaged where anonymous ratings in various categories could be automatically tallied and published, similar to the system used in the Zagat restaurant guide. Then authors who are consistently rated low on organization and comprehensibility might be encouraged to pay more attention to the preparation of their manuscripts. If, in addition, they were consistently rated as unoriginal, limited in interest, and superficial, the author might prefer to take up another line of work. Thus, this system could lead both to an improvement in the quality of papers and to a reduction in their number.

It can be argued that this could lead to the anonymous denigration of an independent worker in a small field dominated by one or two large laboratories. However, a corresponding situation occurs at present in which the worker cannot get published except, perhaps, in very peripheral journals. At least in the proposed system he could present his work for discussion whenever a cognate paper appeared.

Those contributing to the ratings should always have the opportunity to modify them after reconsideration. The use of registration numbers, discussed below, would allow the automatic detection of abuses such as the consistent posting of negative anonymous ratings without any signed commentaries.

On a different level, editors of journals could readily identify papers that are widely recognized as being original, interesting and profound, and could approach the authors for the right to publish them, as previously suggested by Dr. Campanario. The first mechanism of publication identified in your proposal, submission through editorial boards could become unnecessary. In addition, the scores would be of interest to Deans and similar parties who would not have to rely only on the possibly prejudiced assessments from a few senior colleagues.

Instead of requiring approval from two credentialed individuals, a variant of the present form of refereeing, the right to publish a paper should be automatic for anyone with a qualification, a bachelor's degree or similar, from an accredited organization. Every qualified individual would be registered and granted a coded identification number on submission of their first paper to e-biomed. The rare individual who had not passed through the formal system of education but wished to submit an article could be registered by two established scientists as you suggested. To avoid extraneous or outrageous material, a democratic control mechanism could be instituted whereby if six registered members objected to a publication on those grounds, the case could be submitted to a jury of 12 or more other members, chosen randomly, who could recommend that the authors should lose their registration permanently or for a fixed period, or receive a warning, according to the severity of the infraction. To avoid risk to involuntary coauthors, no registered author's name should be allowed on an article unless it was accompanied by his approval. This system could be programmed with other built in safeguards, so that it could be almost self-running and considerably more economical than its alternatives.

The function of the governing body could be reduced to refining these safeguards and adjudicating difficult cases. There will perhaps be an initial period of some confusion but in a few years a body of 'case law' will be built up that will reduce their work load. Actual lawsuits from disciplined or otherwise disgruntled participants could be avoided by requiring them to agree to the regulations of the system at the time of being registered.

David M. Maurice, Ph.D.
Columbia University
College of Physicians & Surgeons
Department of Ophthalmology

Douglas M. Jewett, July 13, 1999

In programmatic approaches to science is may be useful to think in the following way: What will "publication" be like 20 or 50 years from now if the system is simply allowed to develop on its own? How might we intervene now to direct the evolution of the system so that the desirable aspects may emerge in 5 to 10 years, say instead of 50 years? This tends to emphasize the inevitable aspect of change and put it in a positive light.

I think one thing that will change in moving toward a more universal forum is the emphasis of what one makes public. Overwhelmingly, current journals are preoccupied with results. However, the research endeavor itself usually deals more with the formulation of questions, the confrontation with problems emerging in the course of the day, with discrepant observations. All these are of equal validity for the conduct of science, but they are much less public. As it becomes easier to publish, questions may become as important as results in the public forum of science. Curiosity may become as much a public resource as it is now a private attribute. A "result" tends to be attributed to an individual or group, but a good question or problem in the proper forum immediately becomes the property of everyone.

July 12, 1999

Stevan Harnad, University of Southampton, July 12, 1999

On Mon, 12 Jul 1999, Hal Varian wrote:

> This is what strikes me as peculiar in your position. You argue that
> the dramatic change in costs will have a big effect on the academic
> publishing model, and yet the same S/L/P model "is still fine" for the
> trade literature.
>
> Doesn't it seem strange that the same change in costs will have such a big
> effect in one area and a tiny effect in another?

(It's not academic vs. trade publishing, but refereed journal papers vs. everything else: including books -- both academic and trade -- and magazine articles.)

I think it has become clear why this disagreement persists. I can't bring myself to believe that books are really a give-away literature, in the way journal-papers always have been (for their authors).

The data you adduce in support of your position consist of the statistics on how little it is that book authors ACTUALLY make for their texts, on average, and I do not contest that, but I don't think it's the decisive evidence (and NOT just because I am a psychologist, concerned only with authors' hopes!).

I think the fact that books have always been commissioned with contracts specifying royalty payments (even if in most cases the royalties are tiny or never materialize at all), whereas refereed journal articles never have been, is evidence in my favour.

So too, I believe, would be the results of the following 3 thought experiments:

How many (1) book authors would have instead signed contracts that promised them limitless public archiving in perpetuo on the Web, in exchange for signing away all royalty rights? -- compared to (2) journal article authors offered the same proposition? [Prediction: (1) few; (2) all]

How many (1) book authors would be willing to PAY (and how much) to ADD to their existing, royalty-based contracts the right to self-archive in perpetuo? -- compared to (2) journal article authors offered the same proposition? [Prediction: (1) few (and little); (2) most (and surprisingly much)]

How much would book authors demand to be PAID in order to alter their existing, royalty-based contracts, so as to add a clause that this work can NEVER be archived free for all? -- compared to (2) journal article authors offered the same proposition? [Prediction: (1) suprisingly little; (2) (more than it is realistic to imagine anyone could ever afford to offer)]

> I argue that the change
> in costs will also have a dramatic effect in the trade literature with
> lots of free publication of unvetted material (see the Web) with people
> paying for filtering/vetting. Hence the business model for the two
> literatures will tend to become more similar.

I accept that an increasing number of things, including texts, will be (and are being) given away on the Web for free, over and above refereed journal papers!

But we are talking about conditional probability here, not raw probability: What is the probability that an item will be a Web give-away-wannabe, GIVEN that it is a book, vs. the probability that it will be a Web give-away-wannabe, GIVEN that it is a refereed-journal paper?

I predict that the latter will be 100% whereas the former will be much, much lower (and even if we don't add the "in perpetuo" clause, allowing temporary web-archiving for promotional purposes); and when electronic book technology becomes more friendly to linear bed/bath/beach reading, the discrepancy will become bigger, not smaller!

As to modular "vetting" -- I haven't a clear enough case in mind, other than peer review, so I'm not sure what would be on sale here: movie reviews?

> But, as you well know, peer review varies dramatically in its quality.
> There are good journals (which tend to be more selective) and bad journals
> (which tend to be less selective) under the current system. If authors
> pay to get published, there will be a natural tendency to increase the amount
> published and reduce average quality.

Please see prior postings on peer review: The journal quality hierarchy will remain in place. Their respective peers will continue to perform (independently, and for free) as now. High rejection rates will continue to prevail at the top; vanity-press, pay-your-way-in at the bottom. And everyone will continue to know the difference, as they do now.

So the effects you describe will only be felt at the bottom of the hierarchy, where they already are now.

>sh> Heaven forfend! The worst of all possible worlds! You have to pay to
>sh> read AND you have to pay to be published! Insult upon Injury!
>
> But you are paying for different things: authors are paying for
> the hosting, readers are paying for the filtering.

Hosting expenses have not even come up. They are part of the academic infrastructure already. I don't reckon the pennies it costs to archive my papers on my institutional server any more than I reckon the pennies it would cost to archive my personal photo albums there (as many do). Same is true with self-archiving on Los Alamos, E-Biomed or CogPrints. The marginal cost per paper is negligible. Hosting costs are a red herring (for academics).

The ONLY author (institution) costs I have ever mentioned have been those of quality control/certification (QC/C).

I don't believe for a minute in reader-end QC/C costs: QC/C costs are like the Cheshire cat's smile. They are all that is left of journal publication costs once production and dissemination are offloaded on public Archives. It is not at all clear what the READER would be buying if he wanted to buy THAT! It's a service (to the author-institution); its OUTPUT -- the refereed paper -- is already in the public archive for free!

> This makes sense
> since the authors value being published (as you repeatedly emphasize)
> and the readers value selectivity.

I'm afraid this doesn't make sense to me at all! I know what the author-institutions are buying: the vetting and certification of their "product." But what are the reader-institutions supposed to be buying, given that that "product" is archived for free for all?

Yes, readers value selectivity. And it is (in part) because readers value selectivity that authors value the service of QC/C!

> I think that your 1/3 figure for
> refereeing process is about right (I have made the same estimates in my
> papers), so it's the same total amount paid, by essentially the same people,
> in either case.

Yes, but if that same 1/3 amount is paid at the reader-institution end, then only the readers at THAT institution get access, whereas if it is paid at the author-institution end, ALL readers get access (including those at less solvent institutions). And, as we agreed, that enhanced visibility is a benefit to author-institutions! And since we're talking about the same institutions spending the same 1/3 in both cases, it seems to be a choice between spending it the access-blocking vs. the access-enhancing way. What possible advantage can there be to the former? (Please don't revert to the vanity-press argument!)

> And if there are those that don't want to pay for filtering, they are welcome
> to go out and search for the material they want on their own, spending time
> rather than money. But if there are those that want to pay money to have
> the searching and filtering done for them, why should you object?

You keep imagining, basically, a refereed literature, paid by S/L/P, plus a free unrefereed literature, whereas what I am advocating is a refereed literature (first) FREED by self-archiving (of refereed papers) and then, when S/L/P revenue dwindles because users prefer the free archive, QC/C FUNDED up-front out of 1/3 of the S/L/P savings.

Your view is simply reader-centred. You keep seeing the text as the PRODUCT instead of seeing it as like a free ad, with the quality-control/certification being the SERVICE to be paid for.

And as we saw, the only dividend of continuing to see it your way is that the very same 1/3 expenditure, paid by the very same institution, is, on your model, paid for BLOCKING access to everyone outside the author's institution who cannot pay, whereas on my model, it is paid for OPENING access to everyone outside the author's institution who cannot pay.

> My argument is that you seem to think that you can take the existing
> refereed journal literature, reverse the payment system, and have nothing
> else change.
> To a social scientist, this is highly unlikely. The existing payment
> system influences incentives, and changing the payment system will change
> the incentives.

That would certainly be true if referees were paid, but they are not! The payment is just for implementing the refereeing system.

And the implementation is ultimately answerable to objective quality indicators constraining QC/C such as impact factors; so we agree that journals that dropped standards (selected less rigorous referees, for example) would soon be caught out and would lose their place in the quality hierarchy to more rigorously refereed journals. (So much for the vanity-press argument.)

[I am conscious that I come up for a drubbing below over what I said about Learned Inquiry and commercial factors, so let me say more carefully here that QC/C implementation is not "directly" influenced by "1st order" commercial factors -- such as maximizing the number of author-institutions paying the publication fees, or even of maximizing the publication fees themselves; it's more like a museum trying to maintain its standards by collecting only the very best art -- in a system where the best artists also want their art to go into the very best museums. -- Except that they GIVE it to them rather than SELLING it!]

But I am not saying that there may not be some surprises in the transition from S/L/P to up-front QC/C. That's why I think it needs to be planned in advance. See the thread "The Urgent Need to Plan a Stable Transition" in the 1998 file of: http://amsci-forum.amsci.org/archives/september-forum.html

> > The way for a reader to vote is not with his (institution's) S/L/P
> > dollars, but with his eyes, his citations, his refereeing, and his
> > research! This is not commerce we are talking about, but Learned
> > Inquiry.
>
> Hmm. I thought we were talking about the economics of learned inquiry.

You got me on that one. (I'm just a psychologist!)

> I'm not talking about post-publication peer commentary; I'm talking
> about post-publication peer review, just as you describe in the following
> excerpt:

But my point in the prior excerpt was that post-publication "peer review" (which I called, I think quite correctly, peer commentary, because that's what it is!) is not a viable substitute for prepublication peer review!

It's as if all the food is put on the market and then people have to listen to the cacophony of consumers to decide what's safe to eat! I'd rather sell (and buy) my eggs already graded by reliable experts!

> Here, I think, is a difference in our conceptions. You think of the "R"
> [Tag for "Refereed"] and "U" [Tag for "Unrefereed"] as being bound to
> the archive. But that seems rather limited. Why not have different
> organizations providing the "R"s and "U"s with pointers to the
> archive?

See the analogy above. I'd rather have my eggs graded by reliable experts, ONCE. Otherwise who will grade the "experts"?

> This way you get competition for rating services, with the appropriate
> incentives for keeping quality up and price down. Some reviewing
> services may charge for their services and some not; let the users
> decide what they want.

But then who will rate the services for me? And meanwhile, what do I do about treating my ailing grandmother, or just getting eggs for tonight's omelette?

I think the current QC/C, though it could stand some improvement, is good enough so I would be satisfied with freeing the current journal literature SUCH AS IT IS now. QC/C reform schemes can be tested empirically, but for now, I want to stick with the current egg-grading methods...

> Certainly figuring out which institutions can afford journals
> involves looking at how much they are paying now vs how much they would
> have to pay for online access.

Correct, but we have already agreed provisionally on the 1/3 of current S/L/P expenditure figure and we know the bottom line: That SOME won't be able to afford it. But if those who can afford it pay it up-front, instead of as S/L/P, then all the have-nots will get the access anyway, and everyone will be better off.

> > You focus on capturing the available money (via S/L/P), whereas I ask
> > "Why not give it away for free for all, and pay the small remaining cost
> > -- quality control -- out of the S/L/P SAVINGS?"
>
> This is misreading of my claim. I don't have any quarrel with giving
> the writing away for free, and having the authors pay for posting.
> ...The question is who will pay for the filtering?

Posting WHAT? I am talking about posting (self-archiving) the REFEREED (= "filtered") papers. (And there is no cost to speak of for the self-archiving, just for the refereeing QC/C.)

> Just as the authors benefit from being published,
> and will pay to do so, the readers benefit from filtering and will pay
> for high quality filtering.

(Let me first us lay to rest the author "payment" for self-archiving, because there isn't any. )

Now, self-archiving WHAT? Unrefereed preprints? Fine. But why would I want to stop there? What do I want everyone, everywhere to see: just the raw draft, or the accepted, certified final draft? The latter, of course.

Now it's true that readers prefer the final draft too, and it's true that the QC/C will cost 1/3 of current S/L/P and will be paid by my institution either way. But if it is paid up-front, I, the author (and my institution), get the further benefit of infinite reach for my work. Whereas the only advantage to keeping payment on the S/L/P end is some hypothetical one that I can't even quite articulate! (Again, it sounds like no contest to me!)

> Of course, if the authors are willing to pay for the filtering (as you
> assume), there is no need for the readers to pay. But the incentives
> are poor in that system---I think that it would tend to degenerate into
> the vanity press you correctly deride.

I think that is as far as we will get with this one, I'm afraid. We agree on the amount. We agree on who pays (the author's institution either way). But if it is paid up front I claim that it frees the literature for all without loss of quality, whereas you claim that it will compromise quality.

I think this can only be settled empirically. I will continue to try to promote self-archiving (of both unrefereed preprints AND refereed reprints). I expect you will want to be cautioning against the latter...

> In your system, the authors pay the journals to have their papers published.
> The analogy is that the drug companies would pay the FDA to have their
> drugs approved. Do you see a problem with the latter business model?

I actually have no idea how the FDA works; perhaps I shouldn't have mentioned it. It was just meant to be an example of a (reliable?) quality controller. But I do know how peer review works, and I know referees are not paid, and as long as they are not, they are incorruptible (incorruptible by how the implementation of peer review is funded: I am not claiming peer review is perfect, or could not be improved).

Peer review is what it says: Specialised work is reviewed by qualified fellow specialists with no financial incentive. I would have hoped that was how drugs were reviewed too, but of course there is a critical difference there: Drugs (like books, and unlike journal articles) are produced to be SOLD not to be given away. If you want to look to a source of corruption, look there!

Stevan Harnad, University of Southampton, July 12, 1999

On Mon, 12 Jul 1999, Hal Varian wrote:

> You may end up being right that S/L/P is no longer appropriate given
> the change in costs. If it was so inefficient, how could it have
> survived for so long?

The answer couldn't be simpler: Because of the technology and economics of print-on-paper! In that medium, S/L/P was the only viable option if one wanted to be published at all.

In the PostGutenberg, online-only era, for refereed journals, the days of S/L/P are over!

But S/L/P is STILL fine for the trade literature (books, magazines). It's only the anomalous, give-away literature that has been freed at last of the "Faustian Bargain" that held it hostage to S/L/P tolls until now.

>sh> And what about the many countries and institutions that can't afford
>sh> either form of access? (And re-calculate that at least 14,000 times for
>sh> each of the refereed journals in Ulrich's that some institutional
>sh> author's work might appear in.)
>
> And what about countries and institutions that can't afford submission
> fees? In the long run, the same costs have to be paid.

I knew, as I wrote that, that this would be the come-back!

The answer is:

(1) Those disenfranchised institutions are currently NET CONSUMERS of the literature (they aspire only to READ it, if they could only afford it!). They are not net providers (they are not publishing much). They could not afford most of the journals under the S/L/P system. So their researchers had much less basis for publishing anything either, being starved of access to the literature.

In the up-front system, these institutions will simply get a free ride from the NET PROVIDERS (research-active, high publishing-rate institutions), but no one will lose as a result of this. (Stealing my paper to read is a victimless crime in the post-print-photocopy age! Among other things, this is the end of the "Copyright Clearance Center" for the journal literature, which is merely a variant of the "P" in S/L/P.)

It should still average out to less than 1/3 of every institution's prior S/L/P budget being rechanneled toward up-front costs.

And as the institutions that were disenfranchised by S/L/P barriers begin to become more research-active as a result of free access to the literature, their research productivity and income should rise, as should their publication rate, and the resulting revenue available for covering those increasing publication costs. (Research and research-impact revenue should always be ahead of QC/C costs by at least a factor of two, if my < 1/3 figure holds.)

(2) What about institutionally unaffiliated scholars? I think a modest slush fund should be able to cover that minoritarian need quite adequately.

> Your argument is
> that the author-institution pays system covers the costs and allows for
> broader readership, an observation with which I agree. However, there is
> a more subtle issue. An economic system tends to favor those who pay. If
> the authors pay, then the system will lean towards the author's goal
> (getting published) whereas if the readers pay the system will lean
> towards the reader's goal (effective filtering.)

This is the vanity-press argument again. Reply: Peer Review. The peer community will continue to maintain the standards, as always, for free! It is only the IMPLEMENTATION of peer review that needs to be paid for, not referee time/effort. And journal rejection-rates and impact-factors will continue to be the marks of quality (and the magnet for authors), not the money exchanged for implementing peer review!

> I'm not sure which effect is larger. But, of course, there is no
> reason why both sides couldn't pay, if that turned out to be the
> appropriate way to align incentives.

Heaven forfend! The worst of all possible worlds! You have to pay to read AND you have to pay to be published! Insult upon Injury!

> > "It is easy to say what would be the ideal online resource for
> > scholars and scientists: all papers in all fields, systematically
> > interconnected, effortlessly accessible and rationally navigable
> > from any researcher's desk worldwide"
> > http://www.cogsci.soton.ac.uk/~harnad/citation.html
> >
> > As an author, how many potential readers of my work would I like to
> > deprive of this resource -- in the interests of a reader-end S/L/P
> > model (from which I do not make a penny, and which costs my institution
> > at least twice as much as barrier-free QC/C would)?
>
> The publication system shouldn't be designed only to serve authors---it
> has to serve the needs of readers as well (especially if they are the same
> people!). One might add terms like "all meritorious papers, systematically
> evaluated and vetted" to your "ideal online resource". (I realize that
> you acknowledge elsewhere that refereeing is a critical part of academic
> publication, even though it ends up being missing as a desideratum here.)

Precisely. It is and always has been the freeing of the REFEREED JOURNAL LITERATURE to which all these efforts have been directed. And as far as I can tell, that completely nullifies your objection here!

>sh> This "vanity press" model of
>sh> author-pays profoundly misunderstands peer review!
>
>sh> The prestige of the top journals is based on their quality, which
>sh> in turn depends on their quality-control standards: They only
>sh> accept the very best papers (and their typically high citation
>sh> impact factors reflect this). (They are not "designer labels," for
>sh> the patina of which a "consumer" is willing to pay more!)
>
> I think that your subsequent analysis is a more-or-less correct analysis
> of the pressures for quality in the current system. Essentially, low
> quality journals are cancelled since their benefits aren't worth their
> costs. But in your proposed system, the reader bears no costs, so
> this particular feedback is eliminated.

The way for a reader to vote is not with his (institution's) S/L/P dollars, but with his eyes, his citations, his refereeing, and his research! This is not commerce we are talking about, but Learned Inquiry.

> You may well respond that authors will want to submit to quality journals,
> a point I accept. But what does "submit" really mean in this world?
> I have argued elsewhere that when publication costs were expensive,
> it made sense to evaluate ex ante. Now that publication costs are
> cheap, it makes sense to evaluate ex post.

Untested speculations about replacing peer review by post-publication peer commentary are a can of worms on which I've written before:

Excerpt from:
Harnad, S. (1998) The invisible hand of peer review. Nature [online] (5 Nov. 1998) http://helix.nature.com/webmatters/invisible.html
Peer Commentary vs. Peer Review
"And is peer commentary (even if we can settle the vexed "peer" question) really peer review? Will I say publicly about someone who might be refereeing my next grant application or tenure review what I really think are the flaws of his latest raw manuscript? (Should we then be publishing our names alongside our votes in civic elections too, without fear or favour?) Will I put into a public commentary -- alongside who knows how many other such commentaries, to be put to who knows what use by who knows whom -- the time and effort that I would put into a referee report for an editor I know to be turning specifically to me and a few other specialists for our expertise on a specific paper?
"If there is anyone on this planet who is in a position to attest to the functional difference between peer review and peer commentary (Harnad 1982, 1984), it is surely the author of the present article, who has been umpiring a peer-reviewed paper journal of Open Peer Commentary (Behavioral and Brain Sciences http://www.princeton.edu/~harnad/bbs.html, published by Cambridge University Press) for over 2 decades (Harnad 1979), as well as a peer-reviewed online-only journal of Open Peer Commentary (Psycoloquy, sponsored by the American Psychological Association, http://www.princeton.edu/~harnad/psyc.html for what will soon be a decade too).
"Both journals are rigorously refereed; only those papers that have successfully passed through the peer review filter go on to run the gauntlet of open peer commentary, an extremely powerful and important SUPPLEMENT to peer review, but certainly no SUBSTITUTE for it. Indeed, no one but the editor sees [or should have to see] the population of raw, unrefereed submissions, consisting of manuscripts eventually destined to be revised and accepted after peer review, but also (with a journal like BBS, with a 75% rejection rate) many manuscripts not destined to appear in that particular journal at all. Referee reports, some written for my eyes only, all written for at most the author and fellow referees, are nothing like public commentaries for the eyes of the entire learned community, and vice versa. Nor do 75% of the submissions justify soliciting public commentary, or at least not commentary at the BBS level of the hierarchy." http://www.cogsci.soton.ac.uk/~harnad/nature2.html

Food for thought: Would you rather have an ailing relative treated on the basis of the traditionally peer-reviewed biomedical literature, with referees selected and their reports adjudicated by a qualified, answerable Editor, or on the basis of navigating a Netnews chatgroup peppered with "articles" and "comments" by God knows who (guided by hit rates?).

[cf. http://www.bmj.com/cgi/shtml/misc/peer/index.shtml]

> Furthermore, there is
> no reason to use a 0-1, publish/don't publish scale any more---much
> more sophisticated systems could be used.

On this topic, see:

http://amsci-forum.amsci.org/archives/september-forum.html 1999 Thread: Independent scientific publication - Why have journals at all?

Short answer: Peer review is not 0/1, red/green light. It is an interactive, iterative feedback cycle that sometimes leads to a paper that passes the threshold for THAT journal in the hierarchy (everything gets published SOMEWHERE eventually). But referees are a scarce resource, and journal quality is equivalent to referee quality and rigour (and rejection rate).

> One scenario is for public-archiving and self-archiving as the publication
> mechanism and an essentially separate system of cataloging/ranking/peer
> reviewing as the filtering system.

This is already covered by the dichotomy: "U" unrefereed preprint vs. "R" refereed reprint (+ journal name "JX"). BOTH can be self-archived (and suitably tagged).

> The question then is who should pay
> for the peer reviewing? I submit that it may well be the readers, due to
> the incentive effects described above.

No, the readers need merely CHOOSE to search only on items tagged "R" in the free Eprint Archive. The refereeing can be provided by peer review (which ain't broke, hence don't need fixin' -- let alone replacin' by untested alternatives).

> >hv> if an organization "can't afford" access, it is
> >hv> likely an accounting illusion rather than actual lack of money.
> >
>sh> I'd like to see the data for that, not even for all 14K journals in
>sh> Ulrichs, but just, say, the top 6.5K indexed by the ISI. And please
>sh> tell me the figures per journal, per institution, per country.
>
> See Lemberg, Richard, 1996 thesis on costs of digitization, UC Berkeley.
> JSTOR did some calculations with the same conclusion, which are reported
> in part by a speech from Bill Bowen, which, I believe, is available
> on the JSTOR Web site.

That does not answer my question: We are not talking about the costs of digitization, current or retrospective. We are talking about how many institutions/countries can and do afford how many journals!

You focus on capturing the available money (via S/L/P), whereas I ask "Why not give it away for free for all, and pay the small remaining cost -- quality control -- out of the S/L/P SAVINGS?"

If there is no other way to free your intuitions from reader-end market thinking, run your whole argument through on advertisements: Why shouldn't advertisers give their ads only to those who can afford to pay for it?

Answer: Ads are not the right PRODUCT to think of! It is ad companies' SERVICES that advertisers want to pay for. (But before this segues into the vanity-press argument again, note that it's only an analogy; for something closer to a homology, you would have to make it the services of a quality-controller/certifier (the FDA?), and one in which the quality assessment itself is done by independent and incorruptible -- because unpaid! -- assessors [referees]!)

Stevan Harnad, University of Southampton, July 12, 1999

The exchange with Hal Varian has been very interesting, but in the interests of hastening convergence, I will be more telegraphic in my quote/commenting in this round, but first a summary:

The only two substantive issues now are (1) an error (about "author charges") and (2) a disagreement (about who should pay for peer review).

(1) Hal speaks of AUTHOR charges, and I keep speaking of AUTHOR-INSTITUTION charges. The annual costs for quality control/certification (QC/C) (less than 1/3 of the total institutional S/L/P costs for full paper and online publication in the present, obsolete system) will not and should not come from authors' pockets but from (less than 1/3 of) author-institution's annual S/L/P savings from total S/L/P cancellation.

The rest of the (marginally vanishing) costs of periodical publication in the new system are to be borne by centralized self-archiving facilities (Los Alamos, E-biomed, CogPrints, Scholar's Forum), backed up by distributed institutional self-archives, plus the increased offloading of word-processing (and soon tagging/mark-up) onto authoring software (which is on the rise anyway).

(2) Hal acknowledges the trade-off between direct benefits (royalty income) from the sale of texts and indirect benefits ("impact" income). (I put this in crass income terms just for the sake of simplicity.) The trade-off is that charging for access (royalty income) means loss of access to those who can't/won't pay (impact income). (I won't even mention that journal authors don't even get the pennies from the royalty income!)

Yet, despite acknowledging this loss of potential readership (hence indirect revenue) caused by S/L/P barriers, and despite agreeing that self-archiving may even be the way for many books, let alone journals, when it comes to the question of how to recover the much lower residual costs of quality control/certification [QC/C], Hal regresses on the S/L/P trade model, seemingly forgetting both the trade-off and the self-archiving option!

Part of this misunderstanding may revolve around institutional S/L/P, which currently SUBSIDIZES readers (at that lucky institution); Hal contrasts this with PERSONAL (out-of-pocket or out-of-grant) payment of author charges in the proposed system, whereas the most natural way to think of it is simply as rechanneling what is already an institutional subsidy from S/L/P costs to the much lower up-front QC/C costs!

If you must think in terms of who the "consumer" is and how he benefits, the consumer is the author-institution in both cases, the benefit in the former case (reader-institution end S/L/P) being to buy in SOME of the journal literature for the use of its scholars and scientists (to enhance their research impact), the benefit in the latter case (author-institution end QC/C) being to buy in ALL of the journal literature for the use of ALL scholars and scientists (thereby enhancing everyone's potential impact) -- and at less than 1/3 of the price!.

When you speak of retaining S/L/P from the author-institution's point of view, always keep in mind their lost potential impact on the eyes/minds of the many institutions and countries that cannot afford that particular journal...

In fact, the best intuition pump I have found for why charging S/L/P for access makes no sense for the refereed research literature is that it is for exactly the same reason that charging for access to advertisements would make no sense!

Now to (telegraphic) quote/comment mode:

On Mon, 12 Jul 1999, Hal Varian wrote:

> ...You argue that the author should pay for this filtering
> role. Perhaps that is right, but one could also make a case that the
> reader should pay since he is the direct beneficiary of the filtering,
> ...now that the other costs have been driven so low.

But this makes the author and author-institution the LOSER of all those potential eyes/minds (and impact income) -- and PARTICULARLY considering how low those "filtering" (QC/C) costs really are!

> ... the funding agencies pay researchers to produce papers which are sent
> to journals to be evaluated and, in most cases, published, which are
> then purchased by libraries for the free use of the researchers.
> ... it is the role of intermediaries that has got us into the current
> mess; if the authors/researchers faced the true costs of the current
> publication system, they would find a way to reform it quite quickly!

Indeed they would, but it would not be by simply lowering S/L/P barriers, but by eliminating them completely, through author-institution end QC/C payment out of 1/3 of S/P/P savings, plus author self-archiving.

> There is something of a tradeoff, but perhaps less than you think.
> There are ways to vary access and recover costs from customers. Even
> now you can purchase a journal subscription directly or go access
> it "for free" in your library. Here are two different kinds of access,
> one more convenient than the other, and they are priced accordingly.
> So it isn't "access" vs "no access" but different degrees of access.
> The same can be done in the online world; whether this is desirable or
> not is a different question.

And what about the many countries and institutions that can't afford either form of access? (And re-calculate that at least 14,000 times for each of the refereed journals in Ulrich's that some institutional author's work might appear in.)

"It is easy to say what would be the ideal online resource for scholars and scientists: all papers in all fields, systematically interconnected, effortlessly accessible and rationally navigable from any researcher's desk worldwide" http://www.cogsci.soton.ac.uk/~harnad/citation.html

As an author, how many potential readers of my work would I like to deprive of this resource -- in the interests of a reader-end S/L/P model (from which I do not make a penny, and which costs my institution at least twice as much as barrier-free QC/C would)?

And what is the counterpart of personal vs. library access in the desk-top, networked world?

> You argue that author charges could pay for peer review. This maybe
> correct, but I worry about the incentives in such a system. Under
> reader pays, the publisher has an incentive to keep quality high in
> order to attract readers. Under author pays, the publisher has an
> incentive to get as many authors to pay as possible, and other
> mechanisms must be used to maintain quality.

I have already replied to this, in response to Frank Norman at the National Institute for Medical Research. This "vanity press" model of author-pays profoundly misunderstands peer review!

Nothing like this will happen; it is based on a misunderstanding of peer review -- and of what it is that makes the prestigious journals prestigious, and hence makes authors prefer to submit to them rather than elsewhere:
The prestige of the top journals is based on their quality, which in turn depends on their quality-control standards: They only accept the very best papers (and their typically high citation impact factors reflect this). (They are not "designer labels," for the patina of which a "consumer" is willing to pay more!)
The way that high standards of quality are maintained is through rigorous peer review: One cannot BUY success in that process; authors must EARN it (by doing high quality work). Otherwise the prestigious journals would simply lose their prestige (and be replaced by other, more rigorously refereed journals, that recapture their standards, and THEREBY the best papers. [And, no, they will not LOWER their charges to capture to higher-prestige authors either! This sort of market thinking is all based on the wrong, old, reader/consumer-end model: or, to put it another way, the "competition" in this nontrade literature is for high-quality papers, not for author-dollars.]).
On the contrary, it is much lower down in the peer review hierarchy, as one approaches the vanity press, that some abuse of the author-end system is conceivable: Authors may try to buy their way into the pages of low-quality journals when they have failed to earn their way into the high-quality ones. But, frankly, I don't find this at all worrisome! Vanity publications are apparent to everyone; they wear the result (low quality) on their sleeves (and their contents, their authorships, and their impact factors); and such journals already exist today, where the "subsidy" currently comes on the reader-institution end rather than the author-institution end -- everyone knows which ones they are, and that caveat emptor prevails when it comes to deciding whether to read them or rely on what they report.

And don't forget: The peers review for free... QC/C costs are only for IMPLEMENTING refereeing, not for the referees themselves (who, like the author, contribute for free). Vide supra.

> if an organization "can't afford" access, it is
> likely an accounting illusion rather than actual lack of money.

I'd like to see the data for that, not even for all 14K journals in Ulrichs, but just, say, the top 6.5K indexed by the ISI. And please tell me the figures per journal, per institution, per country. As an author/advertiser, I would like to know how many potential customers I am really LOSING if I endorse the SELLING of my papers/ads, instead of having my institution pay up front to GIVE them away to one and all...

> The important principle is that the readers are willing to pay
> something for the filtering services provided by the journal. And it is
> this willingness-to-pay that supports the current business model. You
> could be right that "author pays" is superior for the reasons you cite.
> But my worry is that the economic incentives to provide value to the
> reader (via filtering) are weakened.

Vide supra. Peer review will take care of itself (money is not involved in refereeing); focus on the my POTENTIAL readers whose institutions CANNOT pay...

A.F. (Arne) Smith, July 12, 1999

Dear Dr. Varmus:

E-biomed cannot get here soon enough.
I am old enough to have seen the splintering of biology into an ever-increasing number of little specialties; each with it's own journal and association.
The lack of exchange of information between these specialties has been poisonous to the One Science of biology.
I have seen information which could save lives pass unnoticed past people and organizations which do not know what is outside their area.
When the Human Genome Project "hits the fan" it will only be useful if we can see the systemic effects of genes.
We have enough "splitters"; we need to develop "lumpers".
Is there any help I can give in a political direction? I have been active in conservative politics in NY State for some time.

Best regards,
A.F. (Arne) Smith
Tully, NY

Martin L. Meltz, Ph.D., July 12, 1999

Dr. Varmus and Associates:
I have been following some of the discussion regarding E-biomed, and have had a number of thoughts over the years as to how something appropriate could be proposed in this area.

I am writing to alert you, however, that the assumptions that peer-reviewed articles are always of good technical merit, and that scientists with international renown always and continuously publish articles of good quality, are not correct. (Non-peer reviewed information would be even worse).

I am a Professor of Radiology at the University of Texas Health Science Center at San Antonio, and a member of Standards Coordinating Committee 28 of IEEE. Over the last few years, I have facilitated, with many colleagues around the country, a computerized system to rate and record the technical merit of the already PUBLISHED PEER REVIEWED literature in the area of radiofrequency radiation bioeffects.

The reason is simple; in undertaking considerations of whether or not there is a human health hazard from low level exposures to radio freqency radiation, we must look at the peer reviewed published literature. Unfortunately, some of this literature is of such poor experimental quality, it should never have been published. BUT IT HAS BEEN.

So those of us who consider quality science as being important in making scientific and societal decisions have choosen to take the time to create a system to document whether a given published paper is of good scientific and technical quality, or adequate scientific and technical quality, or of poor scientific and technical quality. Please keep in mind that all of the articles I am talking about did pass through editorial boards; the probelm is the technical competence of many of either the reviews, the articles, or both.

With the system(s) you are proposing, I have great concern about the availability of scientific information, since there is really no control point or system defined (yet) for attesting to the quality of the hypothesis, objectives, methods, data, analysis, and conclusions of a given work.

Since the general public will also have access to the same information, individuals or groups of individuals can use any of the data orconclusion, from either of your prosposed data bases, to claim just about anything.

I urge you to give more attention to these concerns; I believe that there is a solution, but it will take time and effort and collaboration of all of the vested interests to design it properly. If this project moves forward in haste, I believe that there will be much wasted effort, and costs to individuals and organizations far beyond anything that you have yet considered.

Best,
Martin L. Meltz, Ph.D.