Harnad, S. (1996) Implementing Peer Review on the Net: Scientific Quality Control in Scholarly Electronic Journals. In: Peek, R. & Newby, G. (Eds.) Scholarly Publication: The Electronic Frontier. Cambridge MA: MIT Press. Pp. 103-108.

IMPLEMENTING PEER REVIEW ON THE NET: SCIENTIFIC QUALITY CONTROL IN SCHOLARLY ELECTRONIC JOURNALS

Stevan Harnad
Department of Psychology
University of Southampton
Highfield, Southampton
SO17 1BJ UNITED KINGDOM

harnad@ecs.soton.ac.uk
phone: +44 703 592582
fax: +44 703 594597
ftp://princeton.edu/pub/harnad/Harnad/
http://louis.ecs.soton.ac.uk/~harnad/
http://www.princeton.edu/~harnad/
gopher://gopher.princeton.edu:9000/1

ABSTRACT: Electronic networks have made it possible for scholarly periodical publishing to shift from a trade model, in which the author sells his words through the mediation of the expensive and inefficient technology of paper, to a collaborative model, in which the much lower real costs and much broader reach of purely electronic publication are subsidized in advance, by universities, libraries, research publication grants, and the scholarly societies in each specialty. To take advantage of this, paper publishing's traditional quality control mechanism, peer review, will have to be implemented on the Net, thereby recreating the hierarchies of journals that allow authors, readers, and promotion committees to calibrate their judgments rationally -- or as rationally as traditional peer review ever allowed them to do it. The Net also offers the possibility of implementing peer review more efficiently and equitably, and of supplementing it with what is the Net's real revolutionary dimension: interactive publication in the form of open peer commentary on published and ongoing work. Most of this "scholarly skywriting" likewise needs to be constrained by peer review, but there is room on the Net for unrefereed discussion too, both in high-level peer discussion forums to which only qualified specialists in a given field have read/write access, and in the general electronic vanity press.


There is no special problem of scientific quality control that is peculiar to the electronic medium. Scholars criticize and evaluate the work of their peers before it appears formally in print. The system is called "peer review." Like democracy, it has imperfections, but it has no viable alternative, whether on paper or on the electronic airwaves (Harnad 1982, 1986).

TRADE PUBLISHING VS. SCHOLARLY PUBLISHING

Later in this paper I will describe how peer review has been implemented in the refereed electronic journal of which I am the Editor, PSYCOLOQUY, but first I would like to induce some perspectival perestroika in my readers: I was once at a reception with a vice-president of NBC, so I took the opportunity you would all no doubt have liked to take in my place, to chastise him roundly for the low quality of his network's programs. He smiled and asked why I thought he was to blame for that. After all, what did I think the "product" of NBC TV was? I replied that it was TV programs, of course. He shook his head and informed me that it was nothing of the sort: "NBC's product is eyeballs, yours and mine, and we sell them to our advertisers. We're perfectly content to put on the screen whatever it is that will make your eyeballs adhere to it. So you get exactly what you pay for -- with your eye balls."

Well, I don't know if this revelation turned things as much on end for you as it did for me, but I'd like to induce an equally radical shake up in your conception of the real "product" and reward structure of scholarly and scientific publication. First, a disclaimer. What I am about to say does NOT apply to trade publication. Trade publication really works the way I had thought TV did: The product is the author's words. These are sold as print on paper. In order to reach your eyeballs (and pocketbook), the author relies on the mediation of the publisher and his technology for setting word to page and then selling page to reader. The alliance between author and publisher is necessary and mutually beneficial; even in today's era of desk-top publishing, most authors prefer not to be masters of all trades in their finite lifetimes; they leave editing, copy-editing, type-setting, proof-reading, printing, marketing and distribution to the experts.

In this symbiotic relationship between trade author and trade publisher, it is quite natural that the author should transfer the copyright for his words to the publisher, for COPYING is precisely what the author wishes the publisher to do, as many times as possible, to reach all those eyeballs with pocketbooks. And author and publisher share exactly the same interest in protecting that copyright from theft. No one else should be able to sell that author's words as his own, and no one should have access to them without paying for it. These shared interests are clear. If the author were not prepared to transfer copyright to the publisher, the publisher would be investing his time, technology and expertise in a product that was available by alternative means (courtesy of someone else's time, technology and expertise, presumably), and it would be quite reasonable for a publisher to decline to invest in a product without the protection provided by exclusivity (there are exceptions to this in the case of different editions, countries or languages, but such details are not really relevant here). For paper publication DOES require a large investment, even in today's era of desktop publishing. The technology of print is not a trivial one, nor is it cheap.

So although I am sure that trade authors would be happy if there were a way to sell their words to readers without the mediation of an expensive technology (one that, among other things, raises the price of their words relative to what they would have cost if they could be marketed directly, like paintings, perhaps, or performances), the sales-deterrent effect of the add-on cost of the print technology is a small price to pay for the advantages of mass production.

It would accordingly NOT be music to a struggling author's ears to hear that his limited print-run work, which was not selling all that well, was widely available and read in a contraband xerox edition for free (or rather, for the cost of polycopying). Hearing this, the author would be as indignant as his publisher, and both would try to take whatever steps were possible to protect their product from theft, and themselves from loss of rightful revenue.

Now comes the perestroika. Suppose this author were not a writer by trade, but a scholar or scientist, someone whose work is likely to be read by at most a few hundred peers (and usually much fewer) in an entire lifetime. Suppose such a one heard about a similar contraband xerox trade being done in his words and work: What would HIS reaction be? The first response would surely be one of delight (as long as the work was not being attributed to someone else): Why are we scholars if it is not to make a contribution to knowledge and inquiry? And surely that contribution is made only if we are read by and influence the work of our fellow scholars, present and future. The scholar/scientist, in other words, wants to reach his peers' eyeballs so as to influence the contents of their minds; his interest is not in the contents of their pocketbooks.

Upon reflection, however, our scholar/scientist would have to wince and duly report the infraction to his publisher and allow him to take steps to put an end to the illicit trade in his words, because if such infractions were allowed to proliferate unchecked, the very vehicle that carries his words to his peers would be at risk. If contraband trade undermines publishers' investment and revenues, then there will be no primary publication to base the contraband on in the first place.

So this potential conflict of interest between scholar and publisher (the former wanting to maximize eyeballs and minimize any barriers -- such as price-tags -- between them and his work, the latter wanting to recover real costs and make a fair return on his investment, and for expert services rendered) is resolved along the same lines as in the trade publishing model: A financial barrier is erected between word and reader, with the reluctant acquiescence of the author and the justified insistence of the publisher.

This was true in the Gutenberg age, in which the only means of reaching eyeballs was through the mediation of the expensive (and slow and inefficient and unecological) technology of paper: That was the only vehicle in town.

Today, this is no longer true, and although the scholarly community is slow to come to a realization of it, the implications of the PostGutenberg technology of electronic networked communication are truly revolutionary (Harnad 1991). I will illustrate with one seemingly innocent new feature, "anonymous ftp" and with it I will recreate the contraband scenario above, but with a rather different outcome:

THE SUBVERSIVE POTENTIAL OF ANONYMOUS FTP

We again have our fabled scholar/scientist, whose motivation in publishing is to reach the eyes and minds of the highest possible proportion of his relatively small community of peers, as quickly and with as few impediments as possible during his fleeting lifetime, while he is still compos mentis and productive and his work still able to benefit from the interaction. He is accustomed to the fact that his article appears in a journal that is subscribed to by about 1000 libraries and individuals world-wide, that he receives (usually) a dozen to (rarely) a few hundred reprint requests after his article appears (generally the week or two after it appears in Current Contents or some other bibliographic alerting service). Before publication, an unrefereed draft of his manuscript might also have been circulated to a variable number of individuals as a preprint, and an updated final draft may continue to be disseminated in that form after publication (as long as his publisher is willing to turn a blind eye to this limited form of auto-contraband). Then there are the citations (ranging from none, to the median half dozen, to the occasional hundreds -- or even higher in those rarities qualifying as Current Contents "Citation Classics"). This is the small, esoteric world of scholarly/scientific communication among the peers of the realm. (There sometimes appears a second and larger incarnation of the rare paper that is reprinted for educational purposes.)

Enter anonymous ftp ("file transfer protocol" -- a means of retrieving electronic files interactively): The paper chase proceeds at its usual tempo while an alternative means of distributing first preprints and then reprints is implemented electronically: An electronic draft is stored in a "public" electronic archive at the author's institution from which anyone in the world can retrieve it at any time. No more tedious scanning of Current Contents and mailing of reprint request cards by would-be readers, and then costly and time-consuming (but willing) mailing of reprints by authors who would be read: The reader can now retrieve the paper for himself, instantly, and without ever needing to bother the author, from anywhere in the world where the Internet stretches -- which is to say, in principle, from any institution of research or higher learning where a fellow-scholar is likely to be. [1]

Splendid, n'est-ce pas? The author-scholar's yearning is fulfilled: open access to his work for the world peer community. The reader-scholar's needs and hopes are well served: free access to the world scholarly literature (or as free as a login on the Internet is to an institutionally affiliated academic or researcher). And the publisher? Well there's the rub. For, unlike the xerox contraband economy, which has not had its predicted disastrous effects on scholarly publication (apart from whatever role it might have played in raising the prices of scholarly journals), the ftp-contraband economy is a tail that can quickly outgrow the dog by orders of magnitude. Will there be any buyers left to pay the real costs of publication?

That all depends on what we mean by publication, and what the real costs of THAT will turn out to be. Paper publishers currently estimate that electronic publication would cost only 20-30% less than paper publication (e.g., Garson, in press). If that is true, then a loss of the revenues needed to cover the remaining 70-80% because of ftp contraband could well do in the entire scholarly publishing enterprise even if it went totally electronic. In which case there would be nothing in those anonymous ftp archives to retrieve. Is this possible?

I think not. Not only do I think that the true cost of purely electronic publishing would be more like the arithmetic complement of the paper publishers' estimates (which are based largely on how much electronic processing saves in PAPER publication), i.e., SAVINGS of 70-80%, but I also think this will put us over the threshold for an entirely different model of how to recover those costs and create a viable purely electronic scholarly publication system. That would be a scholarly subsidy model, whereby scholars' universities (especially their presses and libraries), learned societies and research publication grants support electronic publications, in place of a trade revenue model. Such a system would reflect more accurately the true motivational structure of scholarly publishing, in which, unlike in trade publishing, authors are willing to PAY to reach their colleagues' eye-balls, rather than the reverse: In physics and mathematics, page charges to the author's institution to offset part of the cost of publication are already a common practice in PAPER publication today. In electronic publication, where these charges would already be so much lower, they seem to be the most natural way to offset ALL of the true expenses of publication that remain. That, however, is not the subject of my paper, so I mention it only in passing. One thing of which I feel confident, however, is that, in line with the real motivation of scholarly publishing, scholars and scientists will NOT accept to have anonymous ftp access blocked by paper publishers invoking copyright. Either a collaborative solution will be reached, with paper publishers retooling themselves to perform those of their services that will still be required in purely electronic publishing, or scholars will simply bolt, and create their own purely electronic publishing systems.

What would they need to do to accomplish this? There will always be a need for expertise in editing, copy-editing, page-composition, graphics and lay-out, and proof-reading. But, most important of all, there has to be a mechanism of quality control -- quality of content, rather than just quality of form, which is what the expertise mentioned so far is concerned with. Publishers formerly furnished this quality control -- or did they? In the case of book publishing, house editors are often scholars who make judgments and recommendations about the content of their authors' manuscripts, though usually in conjunction with other scholars, unaffiliated with the publisher, whom they ask to serve as reviewers, to provide criticism and make recommendations about acceptability and revision, if necessary. In scholarly periodical publishing, the journal editor is usually not directly affiliated with the publisher, and the referees he consults certainly are not. They are us: the author's community of peers. This is why this quality control system is called peer review.

So the ones who monitored and guided the quality of the content of scholarly publishing were always the members of the scholarly community itself. Nor were they PAID for their efforts (in periodical publication, which accounts for the lion's share of the scientific literature and a good portion of the rest of scholarly publishing too; in book publishing they were paid a pittance, but hardly enough to make it worth their while if they were not already resigned to rendering this scholarly community service in any case). So the COST of scholarly quality control certainly cannot be written off as part of the cost of paper publishing: Scholars have been subsidizing this with their free efforts all along. For purely electronic publication, they would simply have to be investing these efforts in another medium.

THE ANARCHIC INITIAL CONDITIONS ON THE NET

And it is indeed another medium, one about which most serious scholars today are still quite wary. Why? I think it is because of the peculiar initial conditions of the making of the new medium, its initial demography, and the style that has become associated with it. Erroneous conclusions about the medium itself have been drawn from these its first messages.

The Net was created, and is continuing to evolve, as the result of a collective, anarchic process among computer programmers ("hackers") and professional, student, and amateur users -- a networked effort, so to speak. Hence it was perfectly natural to imagine that this creative and enterprising anarchic spirit, which has proven so effective in forging these remarkable new tools, should also be the means of deploying them. Indeed, the rapid proliferation of bulletin boards, discussion groups, alerting services and preprint archives, complemented now by simple and powerful search and retrieval tools, all pointed in the direction of a new "ultrademocratic" approach to information production and distribution in this new medium.

Problems immediately manifested themselves, however, in this informational Utopia: Discussions would wax verbose and sometimes abusive; misinformation was difficult to distinguish from information; an ethos of egalitarian dilettantism prevailed; and, worst of all, serious scholars and scientists distanced themselves or kept their distance from the Net, concluding, understandably, that it was much too chaotic and undiscriminating a medium to be entrusted with the communication and preservation of their substantive ideas and findings.

And so things stand today. There are a few brave new electronic journals, but the medium is still widely perceived as unfit for serious scholarship, more like a global graffiti board for trivial pursuit. Yet the remedy is obvious and simple; and, as I have suggested, it is not, nor has it ever been, medium-dependent: The filtering of scholarly and scientific work by some form of quality control has been implicit in paper publication from the outset, yet it is not, and never has been, in any way peculiar to paper.

The scholarly communicative potential of electronic networks is revolutionary. There is only one sector in which the Net will have to be traditional, and that is in the validation of scholarly ideas and findings by peer review. Refereeing can be implemented much more rapidly, equitably and efficiently on the Net, but it cannot be dispensed with, as many naive enthusiasts (who equate it with "censorship") seem to think.

IMPOSING ORDER THROUGH PEER REVIEW

I will now describe how peer review is implemented by PSYCOLOQUY, an international, interdisciplinary electronic journal of open peer commentary in the biobehavioral and cognitive sciences, supported on an experimental basis by the American Psychological Association. PSYCOLOQUY is attempting to provide a model for electronic scholarly periodicals. All contributions are refereed; the journal has an editorial board and draws upon experts in the pertinent subspecialties (psychology, neuroscience, behavioral biology, cognitive science, philosophy, linguistics, and computer science) the world over (Harnad 1990; Garfield 1991; Katz 1991).

In addition to refereed "target articles," PSYCOLOQUY publishes refereed peer commentary on those articles, as well as authors' responses to those commentaries. This form of interactive publication ("scholarly skywriting") represents the revolutionary dimension of the Net in scholarly communication (Harnad 1992), but it too must be implemented under the constraint of peer review.

The objective of those of us who have glimpsed this medium's true potential is to establish on the Net an electronic counterpart of the "prestige" hierarchy among learned paper journals in each discipline. Only then will serious scholars and scientists be ready to entrust their work to them, academic institutions ready to accord that work due credit, and readers able to find their way to it amidst the anarchic background noise.

How is peer review normally implemented, in conventional paper journals? The journal has an Editor and an Editorial Board. With some journals it is the Editor in Chief, with others it is the Editor in consultation with the Board, or with Action Editors, who selects the referees, usually one or two per manuscript, a third or more consulted if a deadlock needs to be broken. The referees advise the Editor(s) by submitting reports (sometimes anonymous, sometimes not) evaluating the manuscript and making recommendations about acceptance/rejection and revision. The reports are advisory rather than binding on the Editor, who makes the actual decision, but a good Editor chooses his referees well and then for the most part trusts them; besides, it is only the very narrow specialty journal whose Editor has the expertise to judge all submissions on his own. The idea of peer review is also to free publication from the domination of any particular individual's preferences, making it answerable to the peer community as a whole -- within the discipline or specialty. (Interdisciplinary journals always have added problems in achieving peer consensus, and indeed, even with specialty journals, referee disagreement rates suggest that consensus is more than one can expect from peer review; nor is it clear that it would be desirable; Harnad 1985).

In the social sciences and humanities, journals pride themselves (and rank their quality) on the magnitude of their rejection rates. Eighty to ninety percent of submissions rejected is not unusual for the most prestigious journals in these fields. Prestige in the physical sciences and mathematics is not associated with such high rejection rates; indeed, they tend to be the reciprocal of social science rates, and biological, medical and engineering periodicals' rates fall somewhere in between (Hargens 1990). In all these fields, however, irrespective of the prevailing rejection rates, there is a prestige hierarchy among journals, with some known to accept only the best work in the field, and some not much more selective than the unrefereed vanity press that exists at the bottom of each field's hierarchy. It is thought that the lower rejection rates in physics may occur because in this field authors exercise more self-selection in choosing which journal to submit their work to, saving only the best for the elite journals. It is also true that in all fields virtually everything that is written gets published somewhere; in the social sciences a manuscript may be submitted to a succession of lower and lower standard journals until it finds its niche; in physics authors may head more directly for something within their reach the first time round.

Another pertinent feature of this hierarchical system of quality control is that most published work is rarely if ever cited. Only a small percentage of what is published is ever heard of again in the literature. This may be because too much is being published, but it may also reflect the inevitable wheat-to-chaff ratio in all human endeavor (Harnad 1986). As a consequence, a scholar is protected on both sides: There is not much risk that a truly valuable piece of work will fail to be published, though it may not make it to its rightful level in the hierarchy, at least not right away. (Peer review is far from infallible.) On the other hand, it is also safe for a scholar, in this monumental information glut, to let the quality control mechanism calibrate his reading, saving it for only the best journals. Again, there is some risk of missing a gem that has inadvertently been triaged too low, but, given the prevailing odds, that risk is itself low.

I have not described a perfect or ideal system here; only the reality of peer review and the reasonably reliable rank-ordering it imposes on scholarly output. It should be apparent that there is nothing about this system that could not be implemented electronically, indeed, there are several ways in which electronic peer review can be made more efficient, fairer, and perhaps even more valid in the electronic medium. The "point faible" of the peer review system is not so much the referee and his human judgment (though that certainly is one of its weaknesses); it is the SELECTION of the referee, a function performed by the Editor. Hence it is really the Editor who is the weak link if he is selecting referees unwisely (or, worse, not heeding their counsel when it is wise). Editors usually have "stables" of referees (an apt if unflattering term describing the workhorse duties this population performs gratis for the sake of the system as a whole) for each specialty; in active areas, however, these populations may be saturated -- a given workhorse may be in the service of numerous stables. So one must turn to less expert or less experienced referees. In practice, the problem is less the saturation of the true population of potentially qualified referees but the saturation of that portion of it that an Editor KNOWS of and is in the repeated habit of consulting.

One of the results of this overuse of the workhorses is that the entire refereeing process is a very sluggish one. One does one's duty, but one does it reluctantly, other duties take priority, manuscripts sit unread for unconscionably long times, referees are delinquent in meeting the deadlines they have agreed to, and sometimes, out of guilt, hasty last-minute reports are composed that do not reflect a careful, conscientious evaluation of the manuscript. There is much muttering about publication delay, a real enough problem, especially in paper publication, but peer review itself is often responsible for as much of the delay as the paper publication and distribution process itself.

Now, as I said, there are no ESSENTIAL differences between paper and electronic media with respect to peer review. And the Net is populated by frail human beings, just as the paper world is. But the Net does offer the possibility of distributing the burdens of peer review more equitably, selecting referees on a broader and more systematic basis (electronic surveys of the literature, citation analysis, even posting Calls for Reviewers to pertinent professional experts' bulletin boards and allowing those who happen to have the time to volunteer themselves). The speed with which a manuscript can be circulated electronically is also an advantage, as is the convenience that many are discovering in reading and commenting on manuscripts exclusively on-screen. All in all, implementing the traditional peer review system purely electronically is not only eminently possible, but is likely to turn out to be optimal, with even paper journal editors preferring to conduct refereeing in the electronic medium (I am certainly doing this more and more with the paper journal I edit).

Once peer review is in place on the Net, once the quality hierarchy has been established, serious scholars will no longer have reason to hesitate to confer their best work to the electronic-only medium. Yet my prediction is that this state of affairs will NOT prove to be the critical factor in drawing the scholarly community onto the Net with their serious work. Much has been said about what the critical "value added" feature of the Net will be that succeeds in winning everyone over. We have spoken of decreased costs, but I think that even my estimate that the true expenses of electronic publication will be only 20-30% of paper publication will not be what does the trick. Decreased publication lags and more equitable refereeing on the Net will also be welcome but still not, I think, the decisive factors. Not even the global access to eyeballs unrestrained by the barriers of subscription cost, xeroxing, mailing or postage, nor the possibility of a (virtually) free world electronic periodical library sitting on every scholar's desk thanks to network links, nor the powerful electronic search and retrieval tools (built on anonymous ftp, archie, wais, gopher, veronica, WorldWideWeb, and their progeny) that will be within everyone's reach -- none of these, remarkable as they are, will be the critical value-added feature that tilts the papyrocentric status quo irreversibly toward the electronic airways.

INTERACTIVE PUBLICATION: "SCHOLARLY SKYWRITING"

The critical factor will be a spin-off of that very anarchy that I said had given the new medium such a bad image in the eyes of serious scholars, what had made it look as if it were just a global graffiti board for trivial pursuit: For once it is safely constrained by peer review, this anarchy will turn into a radically new form of INTERACTIVE PUBLICATION that I have dubbed "Scholarly Skywriting," and this is what I predict will prove to be the invaluable new communicative possibility the Net offers to scholars, the one that paper could never hope to implement.

I think I may be peculiarly well placed to make this prognostication. For over fifteen years I have edited a paper journal specializing in "Open Peer Commentary": BEHAVIORAL AND BRAIN SCIENCES (BBS, published by Cambridge University Press) accepts only articles that report especially significant and controversial work. Once refereed and accepted, these "target" articles are circulated (formerly only as paper preprints, but these days in electronic form as well) to as many as 100 potential commentators across specialties and around the world, who are invited to submit critical commentary, to which the author will respond Harnad 1979, 1984b). Among the criteria referees are asked to use in reviewing manuscripts submitted to BBS is whether open peer discussion and response on that paper would be useful to the scholars in the fields involved (the material must impinge on at least three specialties). Each target article is then copublished with the 20 - 30 (accepted) peer commentaries it elicits, plus the author's Response to the commentaries. These BBS "treatments" have apparently been found useful by the biobehavioral and cognitive science community, because already in its 6th year BBS had the 3rd highest "impact" factor (citation ratio; adjusted: see Drake 1986; Harnad 1984a) among the 1200 journals indexed in the Social Science Citation Index. BBS's pages are in such demand by readers and authors alike that it has (based on an informal survey of authors) one of the highest reprint request rates among scholarly periodicals and, of course, the characteristically high rejection rate for submissions -- attesting as much to the fact that there is more demand for Open Peer Commentary than BBS can fill as to the fact that BBS's quality control standards are high.

Yet BBS has some inescapable limitations, because its tempo is far too slow. Peer review (using 5-8 referees, from 3 or more specialties), is, as usual, a retardant, but even if one starts the clock at the moment a target article is accepted, and even if one allows for the fact that preprints are in the hands of one hundred peers within two weeks of that moment, their commentaries received six weeks after that, the author's response four weeks after that, and then the entire treatment appears in print 4-6 months later, these turnaround times, though perhaps respectable compared to conventional forms of paper publication, are in fact hopelessly slow when compared to the potential SPEED OF THOUGHT.

I have discussed the chronobiology of human communication in more detail elsewhere (Harnad et al. 1976; Harnad 1991). Suffice it to say here that the tempo of a spoken conversation is in the same neighborhood as the speed of thought, whereas weeks, months, or years of lag between messages are not. Whatever ideas could have been generated by minds interacting at biological tempos are forever lost at paper-production tempos. Scholarly Skywriting promises life for more of those potential brainchildren, those ideas born out of scholarly intercourse at skyborne speeds, progeny that would be doomed to still-birth at the earthbound speeds of the paper communication.

I hasten to add -- so as to dispel misunderstandings that have already been voiced in the literature (e.g., Garfield 1991) -- that I am not advocating oral speeds for all scientific "publication." First of all, the time to pass through the filter of peer review already puts some brakes on the speed of interaction. Second, even unmoderated electronic mail correspondence is not as fast as a conversation (nor would it be comfortable if it were -- as anyone who has engaged in real-time e-writing "conversations" can attest). Nor is the goal the undisciplined babbling that we all recognize from "live" symposium transcripts. The goal is something in between: Much faster than paper-mediated interaction, but not as fast or unconstrained as oral dialogue. Moreover, the virtue of "Scholarly Skywriting" is as an available OPTION. Just as not every article is suitable for BBS, not every idea or finding is a candidate for interactive publication. But at last the option is there.

And once you have tasted it (as I have -- e.g., see Hayes et al. 1992), I think you too will be convinced that it adds a revolutionary new dimension to scholarly publication and, even more important, will, I predict, increase individual scholars' productivity by an order of magnitude (all those stillborn ideas that now have a lease on life!).

CREATIVE CHAOS

Let me close by returning to the question of quality control. I have argued that peer review can and should be implemented on the Net, and hierarchically, much as it was in paper, generating a pyramid of periodicals, with the highest quality ones at the top and the unrefereed vanity press at the bottom. This, I have suggested, should allay the apprehensions of scholars who had wrongly inferred that the Net was intrinsically anarchic. But now let me say a few words in praise of the chaotic regions of such a partially constrained system: Sometimes the brakes applied by referees are "unbiological" too: If all of our ideas and findings had to pass through narrow peer scrutiny before they could elicit wider peer feedback, perhaps certain ones of them would still remain stillborn. Within the many possible structures and nonstructures one can implement on a Net, unrefereed discussion, perhaps among a closed group of specialists with read/write privileges (while others have read-only privileges) would be a useful complement to conventional peer review or even to electronic adaptations of BBS-style editor-filtered peer commentary in the form of editor-filtered "skywriting" of the kind BBS's electronic counterpart, PSYCOLOQUY specializes in.

Peer commentary, after all, whether refereed or not, is itself a form of peer review, and hence of quality control (Mahoney 1985). Let us be imaginative in exploring the remarkable possibilities of this brave new medium. My argument here has been on behalf of conventional peer review as the principal means of controlling quality, whether on paper or on the Net, and whether for target articles or commentaries. But once such rigorous, conventional constraints are in place, there is still plenty of room on the net for exploring freer possibilities, and the collective, interactive ones, are especially exciting.

REFERENCES

Drake, R.A. (1986) Citations to articles and commentaries: A reassessment. American Psychologist 41: 324 - 325.

Garfield, E. (1991) Electronic journals and skywriting: A complementary medium for scientific communication? Current Contents 45: 9-11, November 11 1991

Garson, L.R. (in press) Investigations in Electronic Delivery of Chemical Information. To appear in: Proceedings of the International Conference on Refereed Electronic Journals: Toward a Consortium for Networked Publications, Winnipeg, Manitoba, Canada, October 1-2, 1993.

Ginsparg, P. (1994) First Steps Towards Electronic Research Communication. Computers in Physics. (August, American Institute of Physics). 8(4): 390-396. http://xxx.lanl.gov/blurb/

Hargens, L.L. (1990) Variation in journal peer review systems: Possible causes and consequences. Journal of the American Medical Association 263: 1348-1352.

Harnad, S. (1979) Creative disagreement. The Sciences 19: 18 - 20.

Harnad, S. (ed.) (1982) Peer commentary on peer review: A case study in scientific quality control, New York: Cambridge University Press.

Harnad, S. (1984a) Commentary on Garfield: Anthropology journals: What they cite and what cites them. Current Anthropology 25: 521 - 522.

Harnad, S. (1984b) Commentaries, opinions and the growth of scientific knowledge. American Psychologist 39: 1497 - 1498.

Harnad, S. (1985) Rational disagreement in peer review. Science, Technology and Human Values 10: 55 - 62.

Harnad, S. (1986) Policing the Paper Chase. (Review of S. Lock, A difficult balance: Peer review in biomedical publication.) Nature 322: 24 - 5.

Harnad, S. (1990) Scholarly Skywriting and the Prepublication Continuum of Scientific Inquiry. Psychological Science 1: 342 - 343 (reprinted in Current Contents 45: 9-13, November 11 1991). ftp://princeton.edu/pub/harnad/Harnad/harnad90.skywriting

Harnad, S. (1991) Post-Gutenberg Galaxy: The Fourth Revolution in the Means of Production of Knowledge. Public-Access Computer Systems Review 2 (1): 39 - 53 (also reprinted in PACS Annual Review Volume 2 1992; and in R. D. Mason (ed.) Computer Conferencing: The Last Word. Beach Holme Publishers, 1992; and in: M. Strangelove & D. Kovacs: Directory of Electronic Journals, Newsletters, and Academic Discussion Lists (A. Okerson, ed), 2nd edition. Washington, DC, Association of Research Libraries, Office of Scientific & Academic Publishing, 1992). ftp://princeton.edu/pub/harnad/Harnad/harnad91.postgutenberg

Harnad, S. (1992) Interactive Publication: Extending the American Physical Society's Discipline-Specific Model for Electronic Publishing. Serials Review, Special Issue on Economics Models for Electronic Publishing, pp. 58 - 61. ftp://princeton.edu/pub/harnad/Harnad/harnad92.interactivpub

Harnad, S. (1995a) Electronic Scholarly Publication: Quo Vadis? Serials Review 21(1) 70-72 (Reprinted in Managing Information 2(3) 1995)

Harnad, S. (1995b) PostGutenberg Galaxy Wars. Times Higher Education Supplement, Multimedia, p. vi, May 12 1995

Harnad, S. (1995c) Universal FTP Archives for Esoteric Science and Scholarship: A Subversive Proposal. In: Ann Okerson & James O'Donnell (Eds.) Scholarly Journals at the Crossroads; A Subversive Proposal for Electronic Publishing. Washington, DC., Association of Research Libraries, June 1995.

Harnad, S., Steklis, H. D. & Lancaster, J. B. (eds.) (1976) Origins and Evolution of Language and Speech. Annals of the New York Academy of Sciences 280.

Hayes, P., Harnad, S., Perlis, D. & Block, N. (1992) Virtual Symposium on Virtual Mind. Minds and Machines 2: 217-238. ftp://princeton.edu/pub/harnad/Harnad/harnad92.virtualmind

Katz, W. (1991) The ten best magazines of 1990. Library Journal 116: 48 - 51.

Mahoney, M.J. (1985) Open Exchange and Epistemic Progress. American Psychologist 40: 29 - 39.

Odlyzko, A.M. (1995) Tragic loss or good riddance? The impending demise of traditional scholarly journals, International Journal of Human-Computer Studies (formerly International Journal of Man-Machine Studies), to appear. Condensed version to appear in Notices of the Amercan Mathematical Society, January 1995. http://www.research.att.com/~amo/doc/complete.html

FOOTNOTES

1. This Utopian transformation has already taken place in at least one area, high energy physics, where, thanks to Paul Ginsparg at Los Alamos, there is already a global preprint archive, HEP, on which the world high energy physics community relies, both to deposit individual author's own papers and to browse and retrieve everyone else's, to the tune of 35,000 "hits" a day! The project began in August 1991 as an informal electronic preprint distribution list among 100 of Paul's colleagues and it scaled up to encompass virtually the entire high energy physics field (and increasing portions of other areas of physics) within a stunningly short time. (For those who are already in the know about "Uniform Resource Locators" (URLs), the HEP archive can be accessed through WorldWideWeb at http://xxx.lanl.gov/ -- see also Ginsparg 1994 and Odlyzko 1995.)