Arthur Smith <apsmith_at_APS.ORG>:
In another thread I made an assertion about prices which I have now
discovered, at least for our journals (Physical Review) was incorrect.
Going back to 1950, we charged $25/year for a journal of close to 4000
pages - that's $172 in 1998 dollars, or 4.3 cents/page. The per-page price
to libraries varied considerably over the subsequent years (the journal
was growing at times by 20% or more annually) with a low of 1.7 cents/page
(1998 dollars) in 1965, and bouncing back up to about 11 cents/page now
as library subscriptions have declined.
Since we've always been non-profit (ie. revenue equals costs to within
roughly 10 percent) the cost per published page can be estimated by
multiplying the subscription price by the circulation numbers. I don't
have exact numbers for circulation over the years, but I do know the
rough trends going back to about 1960, and the remarkable thing is
this product (price/page * number of library subscriptions) has been
almost constant, probably for the entire 50 years, to within about 25 percent
(except for some anomalous years around 1965 when the price fell behind
the growth in size) - $180/page plus or minus $45, if current circulation
is assumed to be 2000 libraries.
So over at least 40 and perhaps 50 years we have maintained a pretty
much constant cost/page - note that page size and density for our journals
did increase in the 1970's and 80's, so there may have been some improvement
in value per dollar to the community as a whole. But certainly not much.
In this context, even a 30% reduction in costs over the next decade would be
pretty remarkable. I think such a reduction is almost certain (99% probable)
because print distribution will disappear - just as it has already
for the encyclopedia market. There may be a handful of print subscribers
remaining in 10 years, but costs to support those will be minimal, less than
1% of current total costs. Hardware and networking to support electronic
distribution will also be at most 1% of current total costs.
Since printing and print distribution account for 25-30% of current total
costs, we'll probably see at least 25% of costs go away in at most a decade.
Other savings will come from electronic distribution to editors
and referees etc. prior to publication - the paper and mailing for this
amount to at most 2% of total costs right now though so this would not
be a big effect.
Improved authoring tools hold the promise of cutting another 30% off
the per-page costs by obviating a lot of the copy-editing and other
manual labor required in handling manuscripts. Getting these out to
the author community and getting them accepted will take some effort
however - this is where the 10-year time-frame starts to look a bit
optimistic. But I'd give it a 50-50 chance that 90% of authors will be
creating documents that require minimal copy-editing in 10 years. We'll
still have to handle the remaining 10%, but this promises at least another
25% savings in the long run.
Other possibilities for automation could start to tackle the remaining
40% of costs strictly associated with editorial and peer review work.
Assuming continuing 2%/year productivity growth over and above salary
growth rates (possible but not guaranteed) we could perhaps eliminate
a quarter of that 40% in 10 years.
So with some reasonable assumptions about electronic distribution,
improved author behavior (predicated on improved tools for authors),
and general productivity improvement, we could well be looking at
costs a factor of 3 lower than they are today, in ten years. The savings
will have to come from much more than electronic distribution savings
however.
Note that nothing I have said is predicated on an S/SL/PPV scheme for
supporting the journals - I have outlined only our costs associated
with performing peer review to the same level it is currently for our
journals. If we used an xxx-style free pre- and post-publication mechanism
for distribution, this would shave at most the < 1% of costs associated
with hardware, software, and networking to support electronic distribution
from our location - and we'd probably want to be spending at least that
much anyway as a show of support for free distribution.
The fact that our per-page costs have stayed almost constant (in 1998 dollars)
over 50 years while the number of papers handled has grown by almost two
orders of magnitude indicates that the fundamental costs associated with
performing peer review to an acceptable degree have not changed significantly
in all that time. These costs are not going to suddenly evaporate in a year
or two just because of some change in the way journals are funded and
distributed - yes there will be changes, but it will really take one
or more decades to make any substantial difference.
Arthur Smith (apsmith_at_aps.org)
Received on Tue Aug 25 1998 - 19:17:43 BST