close this bookVolume 4: No. 30
View the documentFunding news
View the documentBusiness news
View the documentInternet news
View the documentJob opportunities
View the documentDiscussion groups
View the documentArtificial brains
View the documentScholarly publishing
View the documentComputists' news

An ARL directory lists 50 electronic journals on the Internet, half of them refereed. (In 1991, the numbers were just 27 and 7.) University Libraries' "Report of the Scholarly Communications Task Force" on electronic librarianship is available on http://scholar.lib.vt.edu/reports/SCTF-1994.html, or in PostScript and Acrobat formats from /pub/SCP/reports on scholar.lib.vt.edu. [James Powell (jpowell@vtvm1.cc.vt.edu), VPIEJ-L, 7/25/94.]

Stevan Harnad has been suggesting that learned societies archive and index their members' papers. (Pointers to FTPable papers are also needed, but the academic community moves around quite a bit.) The Physics preprint/e-print archives at http://xxx.lanl.gov/ have over 20K users of 30K articles, with over 35K transactions per day -- I'm not sure what constitutes a transaction -- yet the system is unfunded and unstaffed. Submissions are processed, archived, and indexed automatically, and subscribers are sent digests of abstracts. These primitive systems "have entirely supplanted recognized journals as the primary disseminators of research information in certain fields." Send a "help" or "get blurb" subject line to hep-th@xxx.lanl.gov for more information. [Paul Ginsparg (ginsparg@qfwfq.lanl.gov). serialst and VPIEJ-L, 7/5/94.]

On the other hand, Peter Graham points out that preservation of important literature is what libraries do really well. Other institutions have less experience in long-term archiving subject to short-term budgets and technology changes. [psgraham@gandalf.rutgers.edu, serialst, 7/7/94.]

Are the papers archived by university libraries actually used? "The average article in Science Citation Index is cited by no one and read by not many more." [Stevan Harnad (harnad@princeton.edu), serialst, 7/19/94.]

"With technical advance, one electronic copy of a journal could satisfy the world, and delay would be minimal or nil. So far, the way of paying for such a development has not emerged. Indeed, plenty of people are not sure it needs to be paid for. They believe 'the age of the free lunch' has really arrived." -- Bernard Naylor (b.naylor@soton.ac.uk), LIBER, 7/7/94. [Stevan Harnad, serialst, 7/19/94.]

Scholars are seldom aware of the costs for the publications they use. Andrew Odlyzko notes that the average researcher in mathematics publishes 2-3 papers per year. If $50K/year of salary and benefits is the research component, the cost to students, university donors, and US taxpayers is $20K/paper. A typical math journal accepts half of submissions, and editors and referees appear to donate about 1-2 weeks' time ($4K) per accepted paper. Two reviewing journals (Math. Rev. and Zbl.) each use another $500/paper in unpaid reviewer time. (Paid reviewers could manage it for a much lower salary cost, but subscription prices would be higher.) Publishing costs for a 50KB article vary from $900 to $8,700/paper for different journals, with a median of about $4K/paper -- plus $100/paper for each review journal. (An editorial assistant handling all journal correspondence costs about $100-400 per published paper. Typesetting of mathematical text costs $10-20/page in the US, or $200-400/paper If the author were to do this work separate from original composition, the salary cost might be $1K-3K.) Ignoring the $20K research cost and $5K editing and review costs, a typical scholarly article still costs over $4K to publish. Amer. J. Math. has a paid circulation of 1458 (including libraries), and a typical scholar may skim only a couple of articles. If 20 scholars read each paper in detail, the cost is $200/reading. Even if 200 scholars skim it, the cost is $20 each. "This high cost is likely to doom any pay-per-view efforts in scholarly publications," as scholars object even to copyright fees of even $5-$10. Most of the publishing cost is for the first copy, with additional copies of a whole journal -- Physical Review Letters, in this case -- printed and delivered for about $140/year. This is similar to the cost of photocopying, stapling, and mailing. Electronic delivery eliminates only this small expense, perhaps 15-30% of the publishing cost. However, electronic publishing in the broad sense can trim costs throughout the process. Email reduces correspondence cost, and editing is less expensive when authors provide the typesetting and basic quality control. "Most of the high cost of traditional publishing is caused by the need for communication and cooperation among the many experts involved in the process. With modern technology, doing something is becoming easier than explaining to somebody else what to do." If standards are lowered just a bit, production costs can be kept under $1K/paper -- probably under $500. Uniformity of appearance and style might suffer, but uniformity of papers within one journal volume is of little importance to researchers. Even reference style will be less important when citations are hyperlinks to other documents. Pay-per-view will still be impractical, but site licenses can cover the $500/article production costs. Even lower subscription costs are possible with all-volunteer labor. Quality of peer review is a separate issue from paper vs. electronic publishing. Peers are already handling the review process and are contributing as much work as the publishers. Editing and electronic distribution are very little extra work. [amo@research.att.com, serialst, 7/6/94.]

Lorrin Garson (lrg96@acs.org) of ACS notes that chemical literature is complex, with tables, display math, chemical structures, line art, half-tones, color images, over 500 special characters, and three levels each of subscripts and superscripts. ACS creates journals as an SGML-like database (80-85% of the total cost, for editing, copy-editing, proof-reading, etc.) which is then formatted into print products. Costs for scholarly e-journals must be comparable to those for print journals. ACS has been experimenting with electronic delivery since 1980, including 20K pages per year of supplemental material made available on acsinfo@acs.org. Starting 6/19/94, ACS/CAS made page images of all its chemistry journals available via FTP, FAX, and modem. Journal graphics are being set up as callable objects linked to text, and CD ROMs coming out in 7/94 will have full search and half-tone/gray scale/color printing capabilities. Although ACS is non-profit, its publications arm does make money. Eliminating print publication would have little effect on costs. In fact, costs may increase as the society begins archiving esoteric data collections and the software needed to view them. [harnad@princeton.edu, serialst, 7/5/94.]

The physics community has been publishing equations, graphics, etc., over the networks for nearly a decade. Astrophysicists are even submitting .mpeg animations. Scholarly readers don't particularly want page images -- "photographs of a journal," however presented. They want the information itself. Author-prepared electronic texts often have fewer errors and better graphics than those retypeset by professional publishers. "Those organizations for whom publishing electronic journals will prove more expensive than printing do not have a very bright future." [Paul Ginsparg (ginsparg@qfwfq.lanl.gov), serialst, 7/5/94.]

Stevan Harnad says that authors now provide most of the editing, formatting, and art work, and that their institutions could do much of the copy editing. Adding electronic distribution to a print publication will always cost more, but an ab initio electronic journal can be very cheap. Harnad's PSYCOLOQUY e-journal has a readership of 40K. Costs of production, even with refereeing and quality control, are about $.25/reader/year. He believes that 75% of the costs of traditional journal publication can be saved if scholarly societies take over from commercial publishers. [serialst, 7/5/94.]

(Let each scientist be responsible for his or her own publishing, without mandatory peer review, and publishing costs drop to near zero. Authors, institutions, or societies wanting the imprimatur of peer review can arrange for it themselves. Bibliographic and review publications would become more important, and tracking of revisions would be a problem -- but we could work that out. The mathematics community is not ready for unrefereed (or vanity) publishing, but CS might be. Better one repository for an author's papers than nearly identical versions in seven workshop and conference proceedings, journals, and reprint books. A principal difficulty is that many authors write poorly and need independent copy editing or even substantive editing. Either their institutions or their professional societies should make domain-cognizant editors available, and should perhaps insist on their use.)