Stevan: There are several comments that I would like to make in response to your e-mail to me (11/24/99).
I believe that, in the next 5+ years, article databases (centralized or decentralized) will be a welcome complement to (not replacement of) current personal subscriptions and library collections. Initially, they will replace the 100 million or so copies of articles currently distributed (in the U.S.) through ILL/document delivery, reprints and colleague distributed preprints and photocopies. However, over the long run, these databases may be the basis for real change in the communications processes. I'm still convinced that publishers add value to the communication process, although no one has, as yet, quantified exactly how much they have done so. To add value, some resources/costs must be incurred and, as long as this holds, there must be some way to recover these costs; either through some combination of the "troika" you mention (subscriptions, site-licence, pay-per-view) or some new "paradigm" such as your suggestion that it be paid for by the author-institution up front. In fact, in our book we suggest that this approach ought to be re-examined, although I'm not optimistic. By the way, even if a royalty is charged for access to the current distribution of article separates, the revenue to publishers will not be appreciably increased. We make a big point that pricing (or, if you will, cost recovery) may be the biggest challenge in the future of scholarly publishing. Unfortunately, we could find no "magic bullet." We do make a strong point, however, that any system changes must take into account the effect on scientists' time which has been negatively affected by replacing personal subscriptions with library use. This time dominates the "system" costs and is by far the biggest component of the "price" paid by scientists for the information. With sufficient reading of a journal, for example, it can take less of their time to "receive" journal issues (even in paper) than to sort through or search a large database. The disturbing aspect of spiraling prices is that all participants are losing - scientists spend more of their valuable time obtaining articles, libraries are providing less information at a greater cost to them (and their funders are disillusioned), and publishers have lower circulation (and are getting hammered by detractors). Yet, it appears that the total amount of system resources (and their costs) have not changed much over a 20 year period (on a cost per scientist or cost per reading basis). Note that the "true" system costs must exclude the exchanges of money (i.e., subscription payment); otherwise the system "total costs" would be distorted through duplication. You mention that new costs would be less than 20% of what they are now per article. Such a value (20%) does not hold in all circumstances. For example, it varies substantially by circulation. We have tried to estimate the amount of resources used by publishers (with costs attached to the labor, space, equipment, supplies, etc.). Evidence seems to suggest that per article costs vary in unanticipated ways. The size of the journal (in number of articles or pages) is one variable in which unit costs appear to be low with small journals, rises up to an average sized journal (i.e., dis-economies of scale) and then levels off and, perhaps, drops. A comment by Andrew Odlyzko at a meeting triggered a memory I had of looking into this in the late 1970s. Fritz Machlup sent me some raw data (from his publisher survey) so that I could see if there were economies of scale based on the size of journals. The opposite was observed with small journals having low unit costs and large journals high unit costs (on the average). I was going to pursue it more to find out why, but never did. Another variable related to cost/price is the size of the publisher (i.e., number of journals published). There also seems to be a correlation between price and size of publishers (McCabe) which some attribute to monopolistic pricing (and large profits). Some of the difference may be attributable to the low circulation of journals published by large publishers. This has yet to be determined. However, I suspect a more likely culprit is that overhead tends to rise with an increase in size of labor-intensive service organizations; which is the case in scholarly publishing. One final thought. Any "global archive" database(s) must be accompanied with sound search and retrieval capabilities. Much of the valuable reading takes place outside the author community and much of the reading is of older articles. Furthermore, reading by individuals is extending across disciplines. Much of the reading is of "new" information, not previously known to the reader. I'm not sure if these comments reveal anything to you. However, I think your attempts to deal with "publishing" cost recovery should be welcomed and explored further. The title of our book is: Towards Electronic Journals: Realities for Scientists, Librarians, and Publishers. Carol Tenopir, Professor at the University of Tennessee is the co-author. The text has been put to bed, with the bibliography (over 600 citations), author index and subject index nearly complete. Best regards, Don