APPENDIX 115
Memorandum from Jonathan Cowie
1. Having worked for two decades with learned
and professional body scientific societies that publish, time
of which some 17 years were spent with my having either managerial
or direct responsibility for publishing (including a term on Council
of the Association of Learned and Professional Scientific Publishers),
the Committee may find the following comments of use in their
inquiry. I am currently on policy committees for two learned and
professional scientific societies. However these views are my
own.
Summary
2. (i) There is a difference between
(paper) publishing, electronic (internet) broadcasts, subscriber-driven
dissemination, and author pay-for-driven dissemination. These
differences need to be recognised in any discussion about scientific
publishing or scientific dissemination. Each have their advantages
and disadvantages. There is also the problem of common misconception
as to how these various dimensions inter-relate. The "free
access" landscape is not as simple as some portray.
(ii) There is a problem of success in that
more scientific papers are being published than ever before and
that these are being disseminated in an increasing number of more
specialist dissemination routes. This is putting pressure on conventional
dissemination routes (primarily through journals) and new ways
of paying for dissemination are being realised (such as author
paid for, free public access internet routes).
(iii) The above problems not only impact
on library and potentially research budgets but also on research
assessment that relies on citation and impact factors.
(iv) There are problems with internet dissemination
such as future proofing the technologies used for archival purposes.
(v) There are general problems across all
of scientific dissemination of parallel (duplicate) and salami
(compartmentalising) publishing, maintaining peer review standards,
and historical archiving (separate to "future proofing"
above).
(vi) Government (albeit through its appropriate
agencies) would do well to assess the above problems. One way
of encouraging a best-of-all-worlds scenario would be to develop
a code of practice with regards to dissemination (including library
subscription, researcher-dissemination drivers) as well as the
use of papers in assessment (the latter providing the incentive
for adherence to any best practice code).
PRINCIPAL POINTS
Publishing furthers, reviews, enables assessment
and helps fund the scientific landscape
3. Journal publishing is of major importance
to UK science:
(a) in terms of disseminating advances in
science,
(b) reviewing the scientific landscape and
commenting on scientific applications (including science and society
matters),
(c) for research assessment, and
(d) as a major source of income funding
much of the work undertaken by learned and professional scientific
societies.
Printed journals and e-journal broadcasts are
not synonymous and not comparable: there are differences
4. Printed journals and e-journals are not
synonymous and should not be compared as like-with-like. The process
whereby print journals are produced is "publishing",
whereas that for e-journals on the internet is "broadcasting".
Differences between the two manifest themselves in a variety of
ways including:
(a) Differences in archiving. "Information
on paper can survive for hundred of years; information stored
digitally may not be recoverable this time next week. With seven
million pages of new information added to the worldwide web each
day, the volatility of websites has emerged as an urgent problem,
especially as websites are becoming the version of record for
scientific journals." (BMJ 328:61-2, 2004)
If something similar to the Great Fire of London were to happen
today then we would lose a dozen or so academic libraries: it
would not, though, result in the irretrievable loss of an integral
body of information. However a problem with those e-broadcasting
a journal, say due to virus infection or computing error, could
conceivably remove a body of scientific information from access
worldwide perhaps permanently if this was the sole form of archiving.
Another problem in archiving is "future proofing" broadcasts.
It is possible to read acts of Parliament written on vellum hundreds
of years ago, however even in the short period of time (about
a decade) when internet use has been commonplace in UK universities,
there have been a number of changes in the way in which information
is handled so necessitating in regular software upgrades. Solutions
are being explored. For instance, Stanford University Libraries
has initiated a project called LOCKSS (Lots Of Copies Keeps Stuff
Safe) so that libraries physically store their own copy of the
e-broadcast. (http://locks.stanford.edu)
(b) Differences in accessibility. Accessing
published journals necessitates either having had a subscription
to the journal for its appropriate editions, or visiting an academic
library. There are clear limitations but also advantages. I currently
subscribe to four journals and in the past month have been known
to read these in a variety of places including on the train. However
the advantage of open access e-broadcast journals is that one
can access any edition from any PC connected to the web.
(c) Differences in permanence. This
is related to, but different from, a) archiving (see above). Once
a journal has been published, although not fixed on stone, it
is fixed on paper. Conversely e-broadcasting sees format changes,
web address changes (the basis for "net rot" when hyperlinks
between sites fail to work due to change), and even editorial
change after a period of broadcasting. This last has ethical implications.
The above differences need to be at the heart
of scientific dissemination discussion
5. Given that there is a significant, albeit
currently little recognised, difference between scientific publishing
and e-broadcasting, these need to be included at the heart of
any discussion comparing open access e-broadcasts and conventional
scientific publishing.
Much research is already disseminated both in
print and electronically
6. In part because there are merits to both
scientific publishing and e-broadcasting, and arguably in part
because there is already an established history of scientific
publishing, many disseminating the results of scientific research
do so both in print and electronic form. However free access
to the electronic broadcast is usually restricted to past work;
typically work more than between six months and two years old.
This enables a charge to be collected from those that want access
to recent scientific research.
Disseminating science is costly however it is
done
7. Disseminating the results of scientific
research either through publishing or e-broadcasting, is costly.
The biggest single cost is usually the selection process which
includes: editorial honoraria, selection software and use, editorial
office staff and overheads associated with progressing selection.
These are typically double the post-acceptance editorial costs.
Then there are print and web costs depending on the dissemination
method. (Learned Publishing 16(1):15-20, 2003)
The expansion of science in an increasingly specialised
way is increasing dissemination costs
8. The problem is mostly one of success
in that science in terms of numbers of papers published is increasing.
The past decade has seen more scientists than previously, and
each tends to publish moreone driver of which being research
assessment. Further, as science has grown then so has the number
of specialisms leading to more journals with shorter print runs.
This has led to higher subscriptions which in real-terms have
increased by 150% in the decade from 1990 (Economic Analysis
of Scientific Research Publishing, Wellcome Trust, 2003).
For libraries, whose budgets are broadly constant in real-terms,
this trend is not (indeed patently has not been) sustainable and
subscriptions have been cut. The response from publishers has
varied but many have provided added-value by providing e-broadcast
support to their publications. In some instances this takes the
form of all the work stations at a university or research institution
having access to the electronic version of the journal. However
one response has been to provide a very different e-journal, one
that is open access, or free to the user, with costs being met
by those having articles broadcast.
Costs to authors of free access internet journals
are in four figures
9. If researchers choose to get their work
disseminated by public access internet broadcast, as opposed to
conventional publication, then they will bear the costs. The launch
last autumn of PLOS Biology, a free access internet broadcast
with monthly editions, gives an indication as to what these costs
might be. A figure of US£1,500 was cited (Nature 425:554-556,
2003) as the fee to be charged authors for each published article.
Clearly if a productive researcher was to choose to disseminate
four or five papers a year (not untypical) then the cost presumably
to his or her research budget would be US$6,000-$7,500 dollars
This is not a trivial amount and if reflected across a significant
proportion of the university research population, would have a
noticeable impact on the Science Base. Conversely an annual subscription
to most conventionally published journals is less than this.
The Government funding science has choices as
to how to pay for dissemination
10. If there were a move to public
access internet broadcasts, as opposed to conventional publication
(albeit supported with free access e-broadcasts of more than two-year
old archives), then the Government would be broadly faced with
the following choices:-
i. Allow the efficacy of research budgets
to be eroded as researchers pay for their work to be disseminated
by electronic public access.
ii Provide researchers with new funding
to pay for their work to be disseminated by electronic public
access.
iii Provide researchers with old money
(for example from library budgets) to pay for their work to be
disseminated by electronic public access.
In short, the Government would have to decide
whether or not to provide researchers with finance to pay for
dissemination by electronic public access? And if they chose to
provide this finance then they would have to decide from which
pocket it would be paid. It should be noted that some scientific
disciplines benefit more from one form of dissemination than others.
OTHER ADVANTAGES
TO FREE
ACCESS ELECTRONIC
BROADCASTS
Reduced salami and parallel publishing
11. Because free access journals are paid
for by researchers who have their papers placed on the internet,
the researchers presumably will want to limit the costs they bear.
At the moment with print journals (including those with parallel
e-broadcasts) the subscriber bears these costs and so there is
little financial incentive for researchers to restrict the amount
they attempt to have published. This has led to what is called
"salami" publishing and "parallel" publishing.
Salami publishing is where a research divides up his or her research
into smaller components and produces a paper on each: the research
is sliced as a salami to generate more papers. Parallel publishing
is where a research publishes work in more than one journal. Inventing
an example: if there was research published on aphid ecology then
it could conceivably end up being published by both ecology journals
as well as those relating to horticulture. There are sometimes
good reasons to parallel publishing (such as if the bridge the
papers provide link new areas of science to each other), otherwise
it provides an additional spur to overall academic publishing
cost. However it is known that parallel publishing does take place
within journals related to a single discipline (for example see
paragraph 21(ii) below).
Competition could erode the profits of commercial
publishers, but also perhaps learned society incomes
12. If there was a move for research to
be disseminated freely, electronically, and away from conventional
reader subscription journals then one might imagine that conventional
journal publishers would seek to become economically competitive.
There is some financial slack in journal publishing. Reed Elsevier
is reported (BMJ 328:1-3, 2004) to have made annual profits
of $290 million with margins of nearly 40% on core journals. But
learned Societies who also profit from journals would be hit by
any margin trims and their profits do not go to shareholders but
to fund charitable scholarship and professional activity. Learned
Societies provide UK tax payer funding Government-sponsored research
with synergistic benefits. (Government might note this, especially
in light of its response to the Select Committee's previous report
on learned society funding.)
ADVANTAGES OF
SUBSCRIBER-FUNDED
DISSEMINATION
Subscriber (the scientific community writ large)
driven content is probably preferable to author (individual researcher)
driven content
13. Subscriber-funded dissemination has
the advantage that subscribers drive the academic content, which
in journals is not solely restricted to research papers. Editorials,
topic reviews, symposia reports, book reviews, update articles
and (in some journals) news augment research paper content and
are valuable to readers. If dissemination was driven by the authors
of research papers then one might expect this other content to
be threatened.
COMMON MISCONCEPTIONS
Peer review can be as rigorous irrespective of
the nature of dissemination
14. There is no reason why peer review should
not be as every bit as rigorous in free access electronic dissemination
as it is in conventional journals. Both are costly exercises (see
paragraph 7 and 8).
Free-access internet science will not reduce dissemination
costs as some think
15. Moving from subscriber paid for conventional
publication and internet broadcast to free access internet broadcast
will reduce pressures on the public purse because of libraries'
reduced subscription bills; it will not. As explained above, the
bills would need to be picked up by the researchers.
Interdisciplinary researchers are not, as some
think, penalised by subscriber based dissemination
16. That those working in interdisciplinary
subjects are penalised by subscriber-based dissemination; they
are not. While those working in interdisciplinary areas of science
do draw on a broader base of material, they are not particularly
disadvantaged by this. First, academic libraries stock a range
of journals determined by their users which reflects the entirety
of the university or college and do not segregate access by departmental
budget. Secondly, an increasing number of journals allow free
electronic access to their archives of material more than a couple
of years old. Third, some journals (such as, I believe, those
in the Nature Publishing Group) allow authors the right to post
their published papers own their own, or their department's, web
site (self archiving). Subscription journals do not necessarily
prevent free access and with the advent of DOIs it is possible
to Google search on colleagues' recommendations with ease and
speed.
It is not true that all material conventionally
published cannot be eventually free
17. Access to all material published conventionally
has to be paid for by subscribers. This is not true. As stated
above, increasingly journals (though admittedly not all) allow
free access to their archives (which on an on-going basis are
paid for by past subscribers or advertisers). Secondly,
there are initiatives that allow those academics from less-developed
nations have immediate free access to internet broadcasts of conventionally
published journals. The World Bank and the World Health Organization
HINARI scheme enables those from less-developed nations have free
internet access to the electronic versions of a number of leading
health and medical journals. There is a similar scheme called
AGORA established by the Food and Agricultural Organization and,
again, the World Bank for agricultural and nutrition journals.
Having said this I suspect that the Committee will receive evidence
from respected players who believe that totally free access Worldwide
is the only way forward. There is a certain ethical charm to this
stance but on close inspection one might wonder whether is it
really necessary and does one want researchers to both do the
research and fund their own publication? (Game keepers and poachers.)
I am ethically attracted to keeping these apart. The greatest
calls for free access come to those concerned with molecular biology,
especially with biomedical outputs. This is in part because this
is where much of the growth in science in recent decades has taken
place and secondly because such research tends to be more time
sensitive. Here, waiting six months to a year or two for the research
to become freely available can undermine much of the benefit of
free access but this is not applicable to all disciplines. Yet
again some publishing groups, such as Nature's, do allow
authors to post their own research on their own website, and the
BMJ has maintained a free internet access service since
1998 (admittedly paid for out of an advertising revenue of a magnitude
not commonly available to journals). In the face of those journals
that allow some sort of free access, be it one or more of the
forms cited above, I find it difficult to see what those favouring
free-access journals are getting at other than the fact that not
all journals allow free access to past papers or allow authors
to post their own papers. If conventional journals in the main
did allow either or both of these then the call for free-access
journals would, I'm confident, largely evaporate. (This question
of free access is of course quite separate to that of patenting
(for example, the human genome), even if there are some parallels
and similarity in those calling for free access and fundamental
(as opposed to applied) research outputs being patent free.)
A POSSIBLE WAY
FORWARD
Funding Councils need to monitor the impacts of
free access internet dissemination
18. If free access (researcher-paid) dissemination
takes off then one of the drivers will be research assessment.
Consequently impact factors will be crucially important. Will
these be comparable to subscriber-driven dissemination? Do conventional
journal impact factors equate with those from purely electronic
journals? Were are currently in early days. The Funding Councils
need to monitor this area of citation impact factor and be aware
of the issues that the Institute for Scientific Information (ISI)
are currently considering. However if ISI citation impact factors
follow the established route as they do with new subscriber-driven
journals, it will be a few years before they will denote impact
factors to new public access e-journals. Nonetheless, this time
must not be wasted so that the Government is prepared.
Government agencies need to get the best from
all disseminating modes
19. The Government and its appropriate agencies
could consider what they might do to get the best from all disseminating
worlds: electronic broadcasting, paper publishing, subscriber
driven and free-access (paper author fee driven). (Successive
Governments have held great store by their Foresight initiatives,
so this should be straightforward.)
The Government might devise a code of practice
linked to assessment
20. The Government, or its agencies, might
perhaps encourage a code of practice on how to best disseminate
research and link this to assessment. For example, research assessment
might only use print journals that provide free electronic access
after two years as opposed to those that do not. (Many CABI journals
allow free electronic access to their conventional journals papers
over a year or two old.) The code of practice might also allow
scientific authors in paper journals to retain the right to post
their own work on their own, or their Department's, web site?
(This is the current practice of the Nature Publishing Group.)
By ensuring that the internet benefits are brought to print journals
would mean that author paid for e-broadcasts would have to ensure
that scholarship and other standards are up to those of print
journals to remain competitive in terms of scholarship. This would
improve the case for free-access e-journals being included in
research assessment.
A code of practice might address other on-going
problems
21. A code of practice might also address
other problems such as:
(i) poor peer review (cf. Nature 413:93,
2001) and also insist on authors being blind to referees (even
if many referees think they can guess) at least for the
initial assessment (some referees' first act is to check the author's
past citation record and not referee the research) ;
(ii) salami and parallel publishing through
enforcing legally binding author statementsone survey of
over 20,000 papers in 70 ophthalmic journals detected approximately
1.4% duplication and they may not have caught them all (Nature
421:209)this, if reflected across all of science, not
to mention added to salami redundancy, represents a considerable
cost;
(iii) complete openness in annual reporting
of the journal's or broadcast's own impact factor, paper acceptance
rates, average time to publication etc., a problem more associated
with smaller specialist journals; and
(iv) it would be useful for Government to
consider whether the code might give preference to high impact
research outlets associated (even if through a commercial publisher)
with established learned societies as these generate scholarship
benefits to the UK so effectively enabling the tax-payer's buck
to get more bangs. (The Government has already responded to this
Select Committee's previous report on learned society funding
indicating its support and appreciation of the value of such bodies
but declining financial assistance other than through the existed
(and limited) route via the Royal Society: so the Government has
provided its own raison d'être for exploring support
for quality learned society journals.)
An investigation into how best future proof science
archives is required
22. The Government, or its appropriate agencies,
might wish to see how both print journals and their electronic
counterparts are future-proofing themselves for archive purposes.
Though this will be problematic in that we do not know what future
technology will bring, we do know two things:
(a) that technology will change; and
(b) we are inevitable marching towards the
future.
This investigation needs to have an initial
European (not just UK) focus.
We must not pay twice for dissemination
23. Above all the Government needs to ensure
that it does not pay twice for dissemination of research first
through libraries and second out of researchers' budgets.
THE SELECT
COMMITTEE'S
SPECIFIC QUESTIONS
(a) What impact do publishers' current
policies on pricing and provision of scientific journals, and
(b) particularly "big deal schemes", have on libraries
and the teaching and research communities they serve?
24. (a) Publishers are trying to maintain
profits in the face of dwindling unit returns (see above). Broadly
speaking, those publishers with external corporate or share-holder
drivers tend to be the ones trying to maximize profits more aggressively.
University presses, learned societies, `commercial' publishers
owned by governments and academics, tend to be more sympathetic
to their markets. Libraries are having difficulty in maintaining
their range of journals.
(b) big deals can be good. Learned society
driven big deals (even if in partnership with a commercial publisher)
can tend to cut the cost of bringing a specific specialist discipline
or specialism to a library. However one might imagine a university
finding it cheaper to buy a package of journals from a commercial
publisher that includes material not needed by a university (a
commercial publisher's interests are not confined as learned societies
by specialist area). But I should point out that I do not have
a purchasing librarian's perspective. Nonetheless big deals can
be good for British scientists if brought together by several
publishers internationally. For example, British geologists have
hopes for an on-line aggregate they are developing with six other
US based geological publishers called GeoScience World (GSW)
(reported in Geoscientist 14(2), 2, 2004).
What action should Government, academic institutions
and publishers be taking to promote a competitive market in scientific
publications?
25. Assuming that university libraries are
funded proportionally by the number of researchers and/or lecturers
working at the said establishment then maintaining this per
capita budget in real terms, and ensuring that the researchers'
departments choose from this allowance, should be enough to drive
competition. What would be interesting is to check to how researcher/lecturer
per capita library budgets have fared over the past 20
years. University personnel departments will know how many researchers
and lecturers they have employed (including on short-term contracts)
and university administrators (if not head librarians) should
have records of their library's budgets, so this exercise should
not be unduly arduous. Meanwhile researchers paying for personal
subscriptions out of their own pocket will only do so if they
believe that they are getting value. What Government's could usefully
do is to encourage a code of practice (see paragraphs 20 and 21)
to be associated with research assessment. This might include
free historical access (say after six months or a year), and authors
being allowed to post on their own websites within a few months
of publication (self archiving). It might also include encouraging
participation in a World Bank-type free access for poor countries
scheme and give marginal preference to high impact journals associated
with learned societies (even if jointly published with commercial
publishers) as opposed to solely commercial publishers,.
What are the consequences of increasing numbers
of open-access journals, for example for the operation of the
Research Assessment Exercise and other selection processes? Should
the Government support such a trend and, if so, how?
26. Probably not as much as one might think.
First, as explained above, the academic publishing world is not
clearly divided into open-access and subscriber based journals.
Some subscriber based paper journals have historic free internet
access to papers between six months and a year old, and some also
allow authors to post their own papers on their own university
websites and there are rare examples (such as the BMJ)
of being both conventionally subscriber paid for and free on the
internet. But, I assume the question refers to open access internet
broadcasts, in which case they should have the same citation impact
factors as their conventional counterparts if research assessment
effectiveness is to be maintained.
How effectively are the Legal Deposit Libraries
making available non-print scientific publications to the research
community, and what steps should they be taking in this respect?
27. Pass. Never used them to access information,
only to deposit it.
What impact will trends in academic journal publishing
have on the risks of scientific fraud and malpractice?
28. Recent trends in scientific dissemination
do not yet appear to a have markedly changed trends in scientific
fraud or malpractice. My perception (gained from being a regular
reader of some journals like Nature, Science and the BMJ
that have news columns occasionally featuring news of fraud/malpractice)
is that the concern here lies more with the increasing complexities
and specialist nature of science which is being accompanied by
a commensurate decrease in the number of people capable of discerning
a really clever fraud. True, the increased access we all now have
to scientific outputs means that, for example, there is a broader
pool from which to plagiarize and electronic cutting and pasting
is superficially easy but fraud and malpractice tends to be more
sophisticated than this if it is have a reduced chance of detection.
I am probably more concerned about ill-conceived (and especially
with regard to the statistical analysis of) experiments as one
of the major causes of "malpractice" in the strict meaning
of the word.
IN CONCLUSION
We need to secure the future of the best of all
science-disseminating worlds
29. We need to secure the future of the
best of all science-disseminating worlds: paper publishing, internet
broadcasting, and free access broadcasts. The danger is that we
could end up: paying twice for disseminating science, restricting
library coverage and other access, not future-proofing technology
(so lose past work), and that standards may erode (or fail to
improve). Conversely we can accommodate the increasing number
of specialized papers with free historical access, and free current
access for those prepared to spend a little time surfing. We can
also strive for an improved service by those providing outlets
for research and ensure that the British tax-payer gets the best
deal for science through scholarship synergism.
February 2004
|