Separating Quality-Control Service-Providing from Document-Providing

From: Donald King <dwking_at_umich.edu>
Date: Tue, 30 Nov 1999 20:24:56 +0000

I'm enthusiastic about what you, Andrew Odlyzko and others are doing.
There are two aspects of your recent writing that are particularly
appealing - (1) the notion of authors/institutions paying for the
article QC/C service and (2) making a distinction between the "article
QC/C service" and the "product" of the document. I believe that having
authors/institutions pay for the article QC/C service is sound and may
well be the "magic bullet" that would undo the damages caused by
current pricing practices. However, my analysis of the costs associated
with article QC/C services and with the product yields a variation to
your conclusions. I'll try to explain this in two steps.

One important point is that institutions are currently already "paying"
for nearly all the costs of journal systems: authors' time, most
refereeing, publishers' services through library subscriptions (and
personal subscriptions in many non-academic institutions), library
services, and readers' time. Evidence suggests that the total cost of
authors' time, publisher services and library (and other intermediary)
services are roughly equal, but each is less than one-tenth of the cost
of readers' time. Since institutions are willing to compensate authors
(and referees) and pay for libraries (and publisher services through
libraries), there is no particular reason that they should not directly
purchase the article QC/C services; as you suggest. Yet there are some
strong arguments for them to do so. We have a lot of evidence as to why
authors select the journals to which they submit manuscripts: quality
and prestige of the journal, circulation, make-up of readers, speed of
publication, and so on. However, what is missing is what they are
willing to pay for these journal attributes. By "purchasing" the
article QC/C services, the author/institutions should help balance the
relative "utility" of these attributes (including "fee" for services)
and facilitate a healthy competition among article QC/C service
providers.

A far more important aspect of this policy is that we get out of the
damaging effects of current pricing. We also finesse any arguments made
about the "cost" of article QC/C services. Whether these costs are $300
per article (as quoted for the JHEP) or about $1,500 per article that
we observed doesn't matter, because the author/institutions will choose
service providers by balancing what they charge against the service
attributes they provide. If some have too high a profit or overhead or
are simply inefficient, they will lose out. I suspect that a
"shake-out" of article QC/C services and attributes will result in the
betterment of science. Everyone, including publishers, should welcome
such changes.

Your clear distinction of article QC/C service and product is very
useful and should be kept in mind by everyone when discussing the
future of journals. In our book we refer to the two as (1) making
information content and appropriate information attributes available
and (2) medium distribution/access (your product/paper, but with
digital access included). In so doing we arrive at a different
conclusion. Consider only the distribution/access costs to the article
media provider. The provider cost per hit from a digital database is
surely very small, near zero. However, the same is approximately true
of a well read subscription. The publisher's cost per paper
subscription for reproduction and distribution typically is about $30
to $40 per paper subscription and about $5 to $10 for electronic
distribution (to cover subscription maintenance). Thus, the providers'
cost per reading of subscriptions becomes very low with a sufficient
amount of reading, say over 100, by library users. Thus, low "per hit"
costs are essentially independent of the medium (paper or electronic)
or where it is held. It depends on frequency of reading.

This, to me, is one of the powers of your suggestion that
authors/institutions pay for the article QC/C service. Now
readers/subscribers can choose from among paper subscriptions,
electronic subscriptions, or online access to separate copies of
articles depending on the purpose of reading, frequency of reading,
their own processing and/or access costs, ease of use, and personal
preference of versions. The scientists' time will dominate the access
costs and vary substantially depending on circumstances. It may well be
that open archives or some version of them will be best and ultimately
used exclusively. However, some of our rough cost estimates
(particularly including scientists' time) and combinations of the
factors above suggest that there is a niche for all three -- at least
over the near future.

The point is that we should all support whatever distribution/access
means are best for scientists and the scientific community and this may
not necessarily be only one version or another. If the
author/institution payment for article QC/C services is adapted, this
can result in little risk, but with huge benefits.

I realize that the approach above may appear to be "journal-centric",
but I think that it must be for the time being because of their ability
to match authors and readers and to satisfy both author and reader
service attribute requirements. From a cost standpoint (again) it can
be shown that, with a sufficient level of reading, it is less expensive
to rely on a subscription than to search a large database. Of the
average of 18 journals from which scientists read one or more articles,
most of the journal articles should be accessed from searches, but some
probably should continue to be read "collectively" from subscriptions.
However, this may well change over time, but only with very good search
and retrieval systems and support staff when necessary.

Sincerely,

Don.
Received on Wed Feb 10 1999 - 19:17:43 GMT