Update September 11: the BIS Committee has posted its full report here.
Conclusions and recommendations are available here.
Update March 11: the BIS Committee has posted the full set of evidence here.
Dr. Heather
Morrison
Vancouver,
British Columbia, Canada
hgmorris at sfu dot ca
Business,
Innovation & Skills Committee
February 5,
2013
Re: Business, Innovation & Skills
Committee’s inquiry into Government’s Open Access Policy
1. This is an individual submission
from a scholar specializing in open access and scholarly communication and a
long-time open access advocate. This is a substantially different submission
from the one that I recently submitted to the House of Lord’s Science and
Technology Committee.
Executive Summary
2. Changing the Government’s Open
Access Policy from one intended to support ‘gold’ open access publishing to a straightforward
‘green’ open access policy requiring researchers to deposit works for open
access in a UK-based repository is recommended. This is absolutely necessary to
ensure that the works of UK researchers remain open access and available to UK researchers
and the UK public. The Open Access Policy applies only to UK researchers, not
publishers. A researcher can publish in a fully open access journal that uses
CC-BY, which is then sold to another publisher and converted to toll access.
The steady growth of open access journals, and more recently monographs, over
the past few years illustrates that ‘green’ open access policy is sufficient to
drive growth in open access publishing. Conversely, the policy as written is
highly likely to harm ‘gold’ open access publishing, by inflating prices which
is likely to decrease support for this approach outside the UK.
3. The use of CC licenses for scholarly
works should be considered experimental for the time being. None of the CC
licenses map to any definition of open access. Any of the CC licenses can be
used with toll access works. Some of the arguments used for CC-BY do not bear
careful scrutiny. For example, it is a common belief that CC-BY is needed to
facilitate data and text mining. CC-BY is not necessary, sufficient, or even
desirable for data and text mining. Internet search engines routinely conduct
data and text mining on a massive scale without any need for CC-BY. CC-BY can
be used with works that are not at all suitable for data or text mining, such
as locked-down PDFs. The Attribution element of CC-BY is problematic when a
number of data / text mining sources are combined; data experts recommend CC-0
or public domain, not CC-BY.
4. There are aspects of the CC-BY
license that are problematic for scholarship. CC-BY will often be incompatible
with research ethics and rights of third parties whose work is included in
scholarly works. Permitting the creation of derivatives may open up
possibilities for new ways of speeding the advance of knowledge, but it also
opens up the possibility of introducing errors and damaging the reputations of
scholars by facilitating the creation of poor quality derivatives. These are
just a couple of examples. Much more thought and research would be desirable
before a default license for open access scholarly works is accepted.
5. The vast majority of open access
journals do not use Creative Commons licenses at all, and those that do, do not
always choose CC-BY. Only 11% of the fully open access journals listed in the
DOAJ use the CC-BY license. There is evidence that, given a choice, scholars
prefer to use more restrictive licenses. Recent evidence from Nature’s Scientific Reports found that only 5% of
scholars given a choice between 3 CC licenses chose CC-BY.
6. There are problems with
affordability in scholarly communication in addition to access barriers. It is
important to create a future for scholarly communication that is both open
access and affordable. At the average cost of $188 per journal found by Edgar
& Willinsky in a major survey of journals using OJS, the full cost of
global open access publishing could be supported by the budgets of academic
libraries, at a small fraction of current spend, which could free funds to
support emerging needs such as preservation of electronic information and
support for research data services. At the rate of the $5,000 per article
charged by Elsevier’s Cell Press for “sponsored access”, the costs of the
global scholarly communication system would increase by 16%.
About me
7.
Recently, I completed a doctorate at the Simon Fraser University School
of Communication. My dissertation, Freedom
for scholarship in the internet age https://theses.lib.sfu.ca/thesis/etd7530, reports on research that is highly
relevant to this inquiry, particularly chapter 5, on the economics of
transition to open access, and chapter 3, which includes a substantial section
mapping Creative Commons licenses and open access. I have developed and taught
courses on scholarly communication and open access at the University of British
Columbia’s School of Library, Archival and Information Studies and have
published extensively on the topics of scholarly communication and open access,
including the monograph Scholarly
Communication for Librarians: Chandos, 2009. I have also taught Information
rights for the information age at the SFU School of Communication.
8. I am also a librarian with
more than a decade’s experience, primarily negotiating
purchase of electronic resources at a provincial and
sometimes a national level, through my position as Coordinator at BC Electronic
Library Network.
Detailed comments
9. Open access policy should always
require that researchers deposit work into open access repositories – the
‘green’ approach, and never require that researchers publish in open access
venues such as journals – the ‘gold’ approach.
10. Open access policy should stipulate
that researchers deposit works into UK based open access repositories, such as
institutional repositories. The reason for this stipulation is to ensure that
UK funded research remains open access and remains available to the UK research
community and public. To illustrate why this is necessary, consider the
scenario where a researcher publishes in an open access journal but does not
deposit in a UK based open access archive. The open access journal may cease to
exist or be sold to a publisher that decides to change the model from open to
toll access. Note that policies covering UK funded researchers, by definition,
cover the actions of the researcher, not the publisher.
11. It is not necessary for open access
policy to require publication in ‘gold’ open access journals, because ‘green’
open access policies are more than sufficient to provide incentive for
publishers to adapt and offer ‘gold’ open access journals. Over the past few
years, thanks in large part to the leading-edge ‘green’ open access policies of
the UK Research Councils and similar funding bodies elsewhere, an open access
publishing system has emerged and is growing on a steady basis. There are more
than 8,000 fully open access, scholarly peer-reviewed journals listed in the
Directory of Open Access Journals (DOAJ). The net growth of the DOAJ is a
fairly consistent 3-4 titles per day.
12. It is premature to make any
recommendations about which license is optimal for scholarship. For this reason
it is not advisable to insist that researchers publish using the CC-BY license.
13. One of the reasons it is not
advisable to recommend the CC-BY license is because many of the arguments in
favour of this license are not well thought out. For example, on a superficial
level CC-BY appears to reflect the strong open access of the Budapest Open
Access Initiative definition. However, this superficial resemblance is not
reflected in the legal code. For example, CC-BY does not necessarily mean “free
of charge” which is central to any definition of open access.
14. There is a common misperception that
CC-BY is needed to facilitate text and data mining. CC-BY is not necessary,
sufficient, or even desirable for text and data mining.
15. CC-BY is not necessary for text and
data mining. Internet search engines such as Google conduct text and data
mining on a massive scale, on a continuous basis. This text and data mining is
routinely conducted on works with any of the Creative Commons licenses, or no
license specified, and even web pages that are All Rights Restricted. On the
Internet, the way to note that a web page is not available for text and data mining is to use the norobots.txt
in the web page’s metadata. Otherwise, the default is that text and mining is
the norm.
16. CC-BY is not sufficient to permit
text and data mining. The Creative Commons licenses are a means by which
creators or rights holders can waive certain rights that they have under
copyright. However, the CC licenses do not place any obligations on the
licensor. A CC-BY license can be used on a work that consists of locked-down
image files that are not at all useful for text or data mining. A CC-BY license
can also be used on a website that uses the nonrobots.txt metadata that tells
the web that the page is not available for crawling.
17. CC-BY is not desirable for text and
data mining, because the attribution element is problematic when large numbers
of datasets are combined. Data experts are recommending CC-0 or similar types
of licenses for data for this reason.
18. CC-BY licenses can be problematic
for scholarship.
19. CC-BY as a default for scholarly
works is highly problematic, because CC-BY places no obligations on the
licensor. An open access publisher using the CC-BY license can sell all of
their journals to another entity. There is nothing in the CC-BY license that
obligates the purchaser to continue with the open access model; they are free
to convert all of the journals to toll access. This is one of the reasons I
always recommend that open access policy be for ‘green’ open access archiving.
20. CC-BY licenses will tend to conflict
with research ethics and rights of third parties whose works are included in
scholarly works covered by policy. A CC-BY license grants blanket permission to
use works, including commercial works and making of derivatives, to anyone,
anywhere. This means that a picture of a research subject could be harvested
and included in an image bank to sell for a wide variety of uses, including
advertising. Informed consent in this situation would require explaining to research
subject that if their photo is published under a CC-BY license the consequences
could include such scenarios as having their picture (possibly modified) posted
as part of an ad on a bus.
21. CC-BY licenses, by allowing for
derivatives on a blanket basis without requiring permission, can add inaccuracy
into the scholarly record and/or damage the reputation of scholars,
universities, and the UK education system, if poor quality derivatives are
made.
22. CC-BY licenses, by granting
commercial rights on a blanket basis, permit commercial entities to use the
works of a publisher to compete with the publisher for revenue. For example, a
commercial company could set themselves up to automatically capture new content
created by a journal in order to attract advertising revenue that might
otherwise have gone to the journal. This is a threat to journals, particularly
smaller society journals.
23. The full impact of the Creative
Commons licenses at this point in time is not fully known. Allowing for the
creation of derivatives could open up the potential to increase the speed of
knowledge creation and/or the development of useful new tools and services, or
it could slow down progress by facilitating the creation and dissemination of
poor quality derivatives. For this reason, the use of particular licenses for
scholarship at this point in time should be considered experimental. Use of the
CC licenses should be encouraged, but a particular license should not be
selected as a default, and researchers should not be required to use a
particular license.
24. Most open access journals do not use
Creative Commons licenses at all; those that do use CC licenses do not
necessarily use CC-BY. Only about 11% of the fully open access journals listed
in the DOAJ use CC-BY (see Suber, P. June 2012 SPARC Open Access Newsletter, The Rise of Libre Open Access http://legacy.earlham.edu/~peters/fos/newsletter/06-02-12.htm
25. There is some evidence suggesting
that CC-BY is not the choice of scholars themselves. Nature’s Scientific Reports is a gold open access
journal that provides authors a choice of CC license, affording an unusual
opportunity to observe the CC license choice of scholars when all other
variables are equal, e.g. there is no difference in cost based on the license
choice. As reported by Nature’s Grace Baynes to the GOAL Open Access list on
February 5, 2013, only 5% of authors chose the CC-BY license (from
http://mailman.ecs.soton.ac.uk/pipermail/goal/2013-February/001557.html).
Details:
1 July
2012 to 7 November 2012
Introduced
CC-BY;
Three
license choices available
412
papers accepted
* 37%
were CC BY-NC-SA
* 58%
were CC BY-NC-ND
* 5%
were CC BY
26. The affordability of an open access
scholarly publishing system hinges on the average cost per article. The
majority of open access journals do not charge article processing fees, so it
is important not to confuse average cost per article with the APF approach. In
addition to the access problem, scholarly communication has had an
affordability problem over the past few decades. It is important to address the
affordability problem in the transition to open access.
27. By my calculations, if all of the
world’s scholarly peer-reviewed journal articles were published at the $188 US
average per article found by Edgar & Willinsky in their 2009 survey of more
than 900 journals using Open Journal Systems, the full cost could come from
academic library budgets with cost savings of 96% of academic library budgets
(for details, see chapter 5 of my dissertation). It is important to seek these
savings as academic libraries have many new needs to fill, such as preservation
of electronic information and supporting research data services. On the other
hand, if the average cost were the $5,000 per article charged by Elsevier’s Cell Press for “sponsored access”, this
would increase the cost of the system overall by about 16% - and still not
achieve open access, as sponsored access is not really open access, just
free-to-read from the publisher’s website.
28. The RCUK’s generous block grants for
article processing fees are likely to distort the market by inflating costs for
article processing fees. If this approach were to success in achieving open
access, it would be at the cost of increasing the problem of lack of
affordability of the system. However, I predict that this approach will fail,
as the impact of inflating the costs of article processing fees is very likely
to decrease support for open access publishing outside the UK, thus dooming the
sector the grants are intended to support.
29. I predict that an unintended
consequence of the RCUK block grants for article processing fees will be a
decrease in support for this approach outside the UK as this is likely to
inflate costs. This will decrease the competitiveness of the UK research
system, as it will be stuck with costs that researchers elsewhere do not have
to pay.
Thank you
very much for the opportunity to participate, and for the UK’s leadership in
the area of open access policy.
Heather Morrison, PhD
hgmorris at sfu dot ca
Update Feb 25: the Creative Commons response to this consultation can be found here