[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Revision to Physical Review B data


On Apr 19, 2005, at 7:01 PM, David Stern wrote:

Thanks for the clarifications. I would like to point out two items
which I think need further clarification though:

1) Of the 19 papers for Yale in PRB in 2004, only 6 were authors solely
by Yale researchers. Using a rule of 1/5 if only one author of 5 was
from Yale, I find that only 11.2 authors would have had to pay from
Yale.  This is almost a factor of 2 difference.
Realistically, I don't see universities dividing and sharing these costs
into tiny segments for auditing and overhead reasons. There will need
to be a better way to allocate shared payments from multiple grants. This is an implementation issue, rather than a serious roadblock.
Agreed. But you can't proceed as if every institution will pay the full
amount on every paper. It just weakens your argument. Factors of 2 really
matter here.

Either way, the costs are still significantly and unrealistically higher
than present subscriptions.
Assuming the sole source of funding was library budgets, I would agree. Clearly OA requires a more thoughtful funding model/source linked more
directly to research funding in general (where the costs still remain a
small fraction of overall research expenditures). It is a highly
non-trivial problem to rechannel funds though....

2) My number was $2000 per article. Your article in Info Today stated
the fee for the $850/author per paper. I am confused by your
calculation below in which you seem to have resorted to $850/article. Which is correct?
My article in ONLINE (not Info Today, which is the platform)
My bad - sorry!

stated both the lowest cost determined by a real publisher of $850 and
your estimated cost of $2000 with the more expensive elements of
copy-editing and other unstated benefits.
The $2000 includes some margin for non-paying authors. The main issue is
that you are avoiding discussions about diseconomies of scale. You can't
easily compare a journal that publishes 100 papers a year with one that
publishes 5000+.

Are you stating that there is a difference between per-article and
per-paper cost calculations?
No, not all (I wouldn't even know what that means.) From your article:

"An inexpensive OA journal based within the University of Arizona, with
shared responsibilities among the faculty and the library.  Estimates from
the staff are $850/author per paper. There are no author charges now."

To me this means if there are 2 authors, the paper would cost $1700, if 3
$2550 etc. I am confused by "$850/author per paper" and was asking for

They should be the same thing, as the $850 per-article number assumed
all costs of unaccepted manuscripts are included in the final dollar
amount.  I assume your $2000 figure also included all costs.
It does (including a help desk, human resources dept, benefits, a large
paid editorial staff with supporting clerical staff, etc.). It isn't easy
to review 27,000+ papers per year (all Phys. Rev.).

So I think you still aren't representing the true numbers. It should be
noted that our analysis shows that some (might be most!) large US
research institutions (what APS would call Tier 4 now) actually paid
APS less money (in _unadjusted_ dollars_ even!) for their 2004
subscriptions than their 1998 subscriptions despite the fact that we
now include PROLA, provide full electronic journals and linking, and
publish almost 25% more pages. This is primarily due to cancellation of
multiple paper subscriptions that most large institutions had. So while
a changed model might mean large research institutions pay more than
they are currently paying, it may not mean paying more than what they
historically paid.  One might question whether this "stealth" transfer
of support for journal publishing away from large research institutions
over the last decade was actually fair. It would be useful if Yale went
back and did a retrospective study of what savings they have because of
this kind of consolidation.

It seems that all libraries can find cost savings due to your
technological efficiencies and new methods of manuscript delivery. This is good news, and your organization is leading the way in reducing
Note that the savings comes from the elastic costs associated with
improvement in technology for archiving and distribution. These costs are
already a reasonably small percentage of the per article cost and even if
driven to 0 we would still be left with significant expenses.  The costs
associated with the peer-review process itself are almost entirely human,
inelastic costs that won't diminish (unless the nature of the peer-review
process itself is changed).

However, it is not valid to state that we should always be willing to
pay what we once did.
I don't believe that is what I said. On the other hand, one might be
willing to pay more if there were real tangible benefits (such a global
open access to the research output of your institution).

We always have new costs to allocate, and the pot of funds is not
growing at the rate of inflation, never mind the profit rate of many
publishers. It is our professional responsibility to look for less
expensive distribution models ... as long as we don't recklessly
endanger the necessary peer review aspect of good journal publishing.
No argument there!

The other point worth making is that while the APS journals make a good
benchmark because we budget fairly close to breakeven for our journal
operations, they are not the best example for figuring actual saving in
a global OA model. The OA model only will work when it is applied to
your most expensive (per article) journals, not your cheapest (assuming
you use something like the APS benchmark to limit how much you are
willing to pay in author fees - one would like to believe that
commercial publishers strive for the same level of efficiency and
cost-effectiveness that we have achieved at the APS). The fact that the
OA model only works well if it is applied globally and with payments
really set close to actual costs (and not to cover current publisher
revenue/margins [and appropriately adjusted to handle non-payment by
some authors]) is what makes any transition to an OA model such a
difficult one.

I was using PRB for exactly the reason you state, to demonstrate that
even the best model will be extremely expensive, so imagine what would
happen for journals with (at least) four times the profit margin.  My
assumption was that if we migrate toward a break-even model such as
yours, we would still need a much larger revenue base than can currently
be supplied by the most significant authors and institutional producers.
That is likely true - and the revenue shouldn't come (solely) from library
budgets which don't scale with research funding. A properly funded OA
model (one in which the funding for OA was proportional to research
funding levels at an institution) would scale correctly - right now in the
subscription model you are paying to acquire (less and less) of the global
research output.

Anyway, the bottom line is that I think we are both in agreement that a
naive OA model in which library subscriptions are simple redirected to
cover OA costs won't work. The problem is much more complex than that and
requires a more collaborative effort.


Mark Doyle
Assistant Director,  Journal Information Systems
The American Physical Society