Participants discussed open access (OA) and OA publishing. The Budapest Open Access Initiative (BOAI )—which arose out of a meeting convened in Budapest by the Open Society Institute (OSI) on December 1-2, 2001, to accelerate progress in the international effort to make research articles in all academic fields freely available on the Internet—was discussed, and speakers talked about the OA publishing models of the Public Library of Science and the Proceedings of the National Academy of Sciences (PNAS).
Michael Doyle pointed out how two recent announcements put OA in the center of journal publishing. In July, a cross-party of British politicians called on the U.K. government to make all publicly funded research accessible to everyone, “free of charge on-line,” he said. That same month, the U.S. House of Representatives Committee on Appropriations recommended that all NIH-funded research be made freely available six months after publication.1
A number of speakers commented on the American Chemical Society’s policy on open access. Robert Bovenschulte explained that ACS encourages authors to link their article from their own web sites or their institutions’ web sites to the article on the ACS server. He said that access to the article is free for anyone who reaches the article via this way. However, the number of free accesses per article via this path is limited to 50 during the first year, which is a total reached by hardly any articles according to Bovenschulte. One year after publication, the limit is removed.
However, finding and obtaining free articles in this way, rather than providing free access directly from the ACS publications web site is too restrictive for some. Some participants feel that ACS has been cautious about moving toward free back-files and has engendered membership resistance to it, instead of seeing it as a tool to increase membership value. A suggestion was made that ACS experiment with free back-files, and then reevaluate the matter after one year and charge accordingly. “[Open access] is a train coming down the tracks, and it ain’t going to stop. React to it,” Christopher Reed said. Stephen Berry added that OA is an ongoing experiment or set of experiments and it is ludicrous to think that there is a single solution. “We have to experiment,” Berry said.
Michael Keller, HighWire Press, said that HighWire has a free back-issue program. Publishers in this program offer 770,000 free articles. The number increases at roughly 5,000 articles a week; all of these free articles are in science and medicine.
Martin Blume introduced the American Physical Society’s policy on open access. He said that APS currently allows authors to put the PDF of an article up either on a personal or an institutional web page, if it is linked to the APS abstract. Thus, the article is essentially available free, but at the discretion of the author. According to Blume, this policy may be a first step toward immediate open access after publication, but that APS would like its journals to be totally available without access barriers. At the same time, APS currently has two journals that are open-access. One is Physical Review Special Topics: Accelerators and Beams that is supported by institutional sponsorship. A second one is about to start—Special Topics: Physics Education Research—which will be supported by author charges of about $1000 an article.
Andrea Twiss-Brooks called for self-archiving rights on
More recent information on the NIH Public Access Policy is available on the Internet at http://www.nih.gov/about/publicaccess/index.htm.
personal and institutional web sites. She said this might be a great public relations move for publishers that allow it already, even if all scientists do not actually put their articles up on their web sites, even with permission.
The involvement of the government in open-access plans via the NIH was a red flag to some. Gordon Hammes does not think the government should be in the publication business and does not want to see NIH funds used for this purpose. There is concern about the incompleteness of the NIH database, because half of the research in many journals is not supported by NIH and would not be added to an NIH database.
Some participants questioned the six-month wait and the safety of OA to chemical literature. Peter Gregory said that being six months out of date with research, a scientist might as well not bother doing it. He questioned whether open access to chemical literature for the general public is desirable because it would result in information about explosives, propellants, pyrotechnics, bioweapons, and pharmaceuticals becoming freely available.
Most industrial librarians are really taking a “wait-and-see” approach to OA, according to Lou Ann Di Nallo. Open access might not lead to lower costs, she said. She also called for publishers to adopt a standard OA model.
Workshop participant Philip Barnett pointed out that everyone on all sides of the open access issue (publishers, researchers, librarians, and users of scientific journals) have the same goal, which is the widest possible distribution of the research literature. Yet there is often animosity among the different groups with much us versus them hostility. Complete and open communication and publicity regarding the actual cost of publishing will help reduce this animosity.
THE OPEN-ACCESS MOVEMENT
Open access and OA publishing are not one and the same, Steven Harnad explained. The genesis of the OA movement lies in June 1994, when Harnad posted what he called a “subversive proposal” to the Virginia Polytechnic Institute electronic journals mailing list (VPIEJ-L). Harnad’s proposal sparked a seminal on-line debate. A part of it was later published as a book and immediately became the de facto manifesto of an embryonic OA movement.
Harnad believes that scientists, not publishers, are to blame for the fact that the community does not have OA. The intellectual content of the concept underlying open access has all the intellectual complications of a message of the following sort: “Kids, it is raining outside. Put on your raincoat. That is it. That is the intellectual content,” Harnad said. That is, open access is the raincoat that will protect an author from losing research impact in the current state of restricted journal access. He said the arguments against OA—such as issues with recovering publication costs and copyright protection—are the equivalent of, “the rain is good for you. God meant us to be rained upon. Raincoats won’t protect you. It is illegal to use raincoats. Raincoats will disintegrate …, et cetera, et cetera.”
However, OA is more than free use of all scientific research, Harnad explained. The real second-order dividend of OA, above research impact, is protection from loss of research impact. For Harnad, research impact, is more than mean journal impact factor. It is closer to the dictionary sense of the word, the metaphorical meaning of impact, meaning an action that has consequences. Open access will also facilitate the essence of science, the interactive process—whether it is refereeing, commentary, or some other aspect of the collaborative, self-corrective process.
Harnad described a change in the scientist’s mantra, “Publish or perish.” If research is not published, it might as well be left undone, Harnad said. Research is a public collaboration, an interactive endeavor, which is why it grows and sometimes turns into benefits and applications. Yet the mantra of the scientist today is changing. It is no longer only publish or perish, but incrementally more, Harnad said. It is making research openly accessible to every would-be user on the planet who has access to the Internet.
Open-Access Publishing Versus Archiving
The BOAI defined open access and described two versions of it. Open access is toll-free, on-line, full-text access to the 2.5 million articles published each year in the approximately 24,000 peer-reviewed journals on the planet. The two varieties or two roads to open access are BOAI-1, open-access strategy number one, which is self-archiving, and was identified by Harnad as the “green road.” BOAI-2, the “gold road” to OA, is to publish in journals that will provide free open access for the scientist.
Importantly, an OA journal is not one that has adopted the OA journal cost recovery model, the “author-pays” model, Harnad explained. The majority of the 1,300 open-access journals have not adopted the author-pays cost recovery model. Many of them are conventional journals that have either by principle or as an experiment made the online version of their content accessible toll free for all. This means that only 5 percent of the 2.5 million articles annually can be made OA. Yet, according to Harnad, it also means that the green road to open access might be a better choice right now, because the gold road might take too long. Waiting for the gold option to grow means waiting for it to create or convert and fund 23,000 more OA journals and then persuade the authors of the annual 2.5 million articles to publish in the new OA journals. This seems enormous compared to the one hurdle facing the green road: getting the authors of the 2.5 million articles to self-archive them, said Harnad.
About 92 percent of journals are green, including the “alleged bad guys like Elsevier.” Harnad said he was sure that ACS would go green sooner or later, because it is already green up to 50 reprints. Generally, this means that publishers will not sue authors who self-archive, Harnad believes. For
the journals that state as part of the copyright that the author may not post the article anywhere, Harnad recommends posting the preprint or the preprint and the later correction, a strategy he refers to as preprint-plus-corrigenda; it is possible to add the journal reference after the article is published.
The main purpose of OA is to maximize research impact. Research impact is a measure of the size of the research contribution, Harnad said. He said the journal impact factor is not the first or only measure of impact, or the most sensitive measure. “Impact … is the effect that your research has—not on the desk drawer and not on the pages of the journal in which it appears and not only on the hearts, heads, minds, and work of the researchers who are lucky enough to be at an institution that can afford the journal in which it appears—but on every potential user of that research on the planet,” Harnad said.
To prove this point, Harnad and his colleague Tim Brody, at Southampton University, took 14 million articles from the ISI (Institute for Scientific Information) physics database, from 1992 to 2004, and constructed a software agent (trawler) that searched for articles contained in the physics central archive, both self-archived and non-self-archived.2 The percentage of the overall physics literature that was found to be self-archived grew steadily from 2 percent in 1992 to 10 percent in 2001. His hypothesis proved to be right: Harnad found that self-archived articles have a higher citation impact than those behind a subscription firewall. This fact might move funding agencies and institutions toward self-archiving, he said.
Harnad explained that there is a tagging system is in place to help find articles. The 1999 Santa Fe Convention, the Open Archives Initiative (OAI) interoperability protocol, and OAI tagging all influenced the establishment of a system that ensures that every article no matter where it is, is tagged with information such as author’s name and journal name, and are then drawn together into a searchable archive. According to Harnad, the self-archive cycle is as follows: An article is written and should be self-archived (but this is optional) and submitted for peer review. The peer-review cycle takes place. The final refereed version is self-archived and then the new research cycles begin.
Open access archives are not set up to be permanent, Harnad explained. OAI-compliant archives take full text in PDF, HTML, and XML format, among others. Self-archiving is not about archiving in the preservation sense of archiving, so authors do not have to worry about choosing a format for immortality. Harnad has created a search tool called ParaCite, which is being developed to locate articles from raw references using a combination of search engines including Google, OAIster, D-Lib, and others. Right now, the only way to search for the self-archive articles is with Google, Harnad said, but the overall problem is not missing search engines but rather missing content.
There is a competitive advantage in OA for research departments. A department that is completely OA would have an advantage over another department with equal quality articles, because even this little bit of an edge is enough to give it higher impact ranking.
However, the work involved in self-archiving might deter some scientists from making the effort. Christopher Reed thought that after the effort of publishing and keeping a web site, keeping an archive might be too much of an effort for academic researchers. Stevan Harnad said that the institution could take care of archiving, because it gained from the added impact.
OPEN-ACCESS PUBLISHING MODELS
Public Library of Science
Vivian Siegel described the Public Library of Science as a public charity with a mission to make the world’s scientific and medical literature a freely available public resource. A way to achieve this mission is to launch open-access journals. PLoS Biology was launched in 2003, PLoS Medicine in the fall of 2004. Publication is the final and often the only tangible product of research. In an electronic era, it is possible to think about publishing as service providing and assign a fixed cost to the value that publishers add.
It is in the interest of funding agencies and institutions to ensure that the final product of research, a published manuscript, is available to everyone, Siegel said. PLoS helps cover those costs, in part through grants. The PLoS definition of open access is free and unrestricted on-line access to the research literature. PLoS has highly permissive usage licenses. Authors retain copyright, but sign a license. PLoS uses the Creative Commons attributions license. Finally, the papers are deposited in public databases.
PLoS Biology has gotten more than 1 million COUNTER compliant downloads of articles this year, about 100,000 downloads every month, and about 4 million hits each month. PLoS publishes about 20 papers every month. PLoS Medicine had 30,000 COUNTER compliant downloads in the first week of its existence. These numbers do not include downloads at PubMed Central. PLoS also reaches areas that do not have on-line access, Siegel added. There have been cases of the full PDF of the journal being downloaded (by someone other than the author or publisher) and sent to places where quick electronic access is not feasible.
Siegel was asked if PLoS Chemistry or PLoS Physics are currently being planned. She responded that PLoS is currently focused on biology and medicine, and that 2005 does not include a plan for a chemistry journal. “I hope that is enough of a nudge to the existing chemical journals to get
S. Harnad and T. Brody, “Comparing the Impact of Open Access (OA) vs. Non-OA Articles in the Same Journals,” D-Lib Magazine 10(6), (June 2004), accessed on the Internet at http://www.dlib.org/dlib/june04/harnad/06harnad.html.
their act together, before we decide that chemistry is just too slow and too unwilling to adopt these sorts of changes, and that we need to launch an alternative for the chemistry community as well,” Siegel said
In response to the idea of PLoS Physics, Siegel added that physics is a very interesting example because the physics community has been effective at using its preprint server and physics archive. According to Siegel, the kind of sharing of information that already happens in the physical community at a different level leaves much to emulate.
Martin Blume expressed relief by this statement, to which Siegel replied, “I love the fact that [PLoS is] an organization of 25 people, and that I can say something like that and relieve Marty Blume, who has a much larger operation.”
Proceedings of the National Academy of Sciences
Bridget Coughlin described the PNAS open-access experiment, in which authors have the option to pay $1,000 to have their papers be open access at the time of publication.3PNAS is the official journal of the U. S. National Academy of Sciences, which is a private, nonprofit, non-governmental, and self-perpetuating society. However, PNAS does not receive funds from the Academy or from membership dues like other societies, nor does it give back to support Academy activities. PNAS has to budget to zero, utilizing two revenue streams: author charges of $1500 per article on average, and subscriptions, which range from a nominal $250 to just over $6,600. Four percent of what is published is in physical sciences; 8,500 papers are submitted a year, and 1 in 6 is accepted. The archive is free after six months.
Over a period of 10 weeks in 2003, PNAS surveyed 610 corresponding authors,4 Coughlin explained. The journal asked if authors would be willing to pay a surcharge for their articles to be freely available on-line at the time of publication. Of the 210 authors that replied, 49.5 percent said yes, and 50.5 percent said no. The second question was what the maximum amount would be that they would be willing to pay. Most people, about 80 percent, were willing to pay the lowest denominator amount of $500; about 15 percent were willing to pay $1,000 per article. PNAS then polled its editorial board for this OA experiment. Of 110 board members, 22 percent responded; 84 percent said “try the experiment.” PNAS also had the unanimous support of its Committee on Publications.
Starting in the spring of 2004, PNAS asked authors, after acceptance of their papers, if they would pay the $1,000 surcharge for their articles to be freely available on-line at the time of publication, in addition to their page charges. Only eight OA fees were waived, or 4.5 percent of all OA articles. The first article went up in May of 2004 (Figure 7.1). The last few issues (September-October 2004 time frame) were around 15 percent open access. This indicates that authors are “voting with their feet,” Coughlin said. She believes the authors submitted to the journal as a sign of support for the open-access movement. The 15 percent OA research articles are multidisciplinary, with genetics and evolution leading, followed by geology and environmental sciences.
She added that subscribers now obtain access to all articles right away, so that paying the $1,000 fee was truly an extra cost. In 2005, PNAS is planning to adjust the open-access fee to $750 from $1,000, for those from an institution that has a site license, to reduce the burden on institution at large. PNAS is tracking the decay curves (Figure 7.2) of open-access articles versus non-open-access articles, to see if indeed there is traffic without a subscription block, but the numbers are still very small.