3.1

The “Green road” to OA

Green OA was initially defined as authors’ self-archiving of their articles in open repositories . However, with the recent development of Internet and OA mandates, articles can now be archived on authors’ personal websites, or in open repositories by institutions/publishers on the authors’ behalf. For instance, many publishers have been uploading articles funded by the National Institutes of Health (NIH) to the PubMed Central (PMC) on behalf of authors . Therefore, a more general definition of Green OA as used by Björk et al. may be more appropriate:

“All freely accessible copies of articles, including different versions of said articles, which exist on other web locations than the original publisher ’ s website.”

The locations of self-archived articles can generally be categorised into subject-based repositories (e.g. the PMC), institutional repositories (IRs) and other websites (e.g. personal websites/academic social networks, departmental websites, industrial websites). Among these, subject repositories and institutional repositories are usually better locations for self-archiving, because they are maintained by professional librarians, with relevant resources and technologies to ensure the visibility and persistence of archived articles . One example of such mechanisms to improve the visibility of Green OA copies is the OAIster database, a combined catalogue harvesting data from all OAI (Open Archives Initiative) compliant repositories/libraries, which as of 2017 includes over 50 million records . The Open DOAR (Directory of Open Access Repositories; www.opendoar.org ) and ROAR (Registry of Open Access Repositories; roar.eprints.org) are two websites providing information about existing open repositories. According to the Open DOAR, there has been a substantial increase in the number of registered open repositories worldwide, from 902 in 2007 to 3311 in 2016 .

To protect their subscription revenue, publishers of subscription-based journals usually only allow authors to self-archive certain versions of their articles, after a set embargo period. Based on articles’ content and formats, they are usually categorised into three distinct versions:

• •

  Publisher’s version (after copy-editing; also known as the “published article” or “exact copy”);

• •

  Author’s postprint manuscript (before copy-editing but after peer-review; also referred to as “accepted manuscript”); and

• •

  Author’s preprint manuscript (before peer-review; also known as “submitted manuscript”).

SHERPA/RoMEO (Securing a Hybrid Environment for Research Preservation and Access/Rights Metadata for Open Archiving; www.sherpa.ac.uk/romeo ) is a database of the self-archiving policies of journals and publishers. As of April 2017, 80% of all 2362 publishers included in this database allow some form of self-archiving . Miguel et al. studied journals registered in Scopus as of April 2010, and found that 57% of journals in “medicine” and 52% of those in “areas related to medicine” were subscription journals that did not allow any form of self-archiving. However, in another 2010 analysis by output volume , the self-archiving of preprints and postprints was allowed in 79% and 78%, respectively, of articles published in subscription-based health sciences journals. As a result, at least about 80% of all research articles published in health sciences journals could be made OA through the Green route alone.

However, the self-archiving of potentially inappropriate versions of articles (versions not allowed by the publishers) has been noticed in previous studies on OA across scientific fields and more specifically in the field of dentistry . In a 2017 study, Jamali found that among subscription-based articles (not published in Gold OA) self-archived on the ResearchGate, about 50% did not comply with publishers’ policy and therefore infringed the copyright, of which 98% were due to the self-archiving of the publisher’s version PDFs. Researchers need to be more familiar with publisher’s copyright and self-archiving policies and relevant resources (e.g. the SHERPA/RoMEO website) before making their articles Green OA.

3.2

The “Gold road” to OA

The Gold OA, sometimes also referred to as “publisher-provided OA”, is broadly defined as making articles OA through journals or publishers. It can be further subdivided to three types according to the OA policy of journals: “Direct OA”, “Hybrid OA” and “Delayed OA” .

• •

  “Direct OA” refers to publishing in full immediate OA journals (OAJs), which make all their articles freely available online upon publication (e.g. PLoS One , BMC Oral Health ). To cover costs related to editorial workflow, a growing proportion of OAJs require authors (or their funders/institutions) to pay an “article processing charge (APC)” . In 2010, the average value of such APCs was about 1100 USD for OAJs in biomedicine ; among authors who published in OAJs in the fields of health sciences, biology and life sciences, an estimated 80% had their APCs covered by funders or institutions, 8% used their own funds, 8% had their APCs waived, and the rest paid the APCs through other means . According to Laakso et al. , the number and volume of OAJs have been growing rapidly. Between 2000 and 2009, the average annual growth rate was 18% for number of journals and 30% for the number of published articles; in 2009, an estimated 7.7% of all peer-reviewed articles were published in OAJs. The DOAJ (Directory of Open Access Journals; www.doaj.org ) is a major index of legitimate OAJs. As of April 2017, a total of 9358 OAJs are listed in this database, which include 899 medical journals and 93 dental journals .

• •

  “Hybrid OA” means those subscription journals that allow authors to make their individual articles OA by paying an optional fee, while keeping the rest of the journals’ content available to subscribers only (e.g. Journal of Dentistry , Oral Diseases , Clinical Oral Investigations ). After Springer first initiated its “Open Choice” Hybrid OA model (APC: 3000 USD) in 2004 , many major publishers followed suit. Based on a study of the five largest publishers, Laakso and Björk found that the number of Hybrid OA articles published by these publishers increased substantially from 666 in 2007 to 13994 in 2013; among all articles published in dental journals that had at least one Hybrid OA article in 2013, an estimated 3.7% were published using Hybrid OA. Interestingly, although Hybrid OA journals usually have higher APCs than full OAJs, recent studies of authors’ APC payment behaviour have indicated a current preference among authors for Hybrid OA journals over OAJs .

• •

  “Delayed OA” refers to another variant of subscription journals, which keep their latest articles accessible only to subscribers, but make most or all articles automatically OA at the expiry of a set embargo period (e.g. Operative Dentistry , The European Journal of Orthodontics ). As there has been no comprehensive index/database for such journals, they were ignored in most previous studies . In a recent explorative analysis of 492 “Delayed OA” journals, Laakso and Björk found that they were substantial in volume and populated with high-impact journals (e.g. Science , New England Journal of Medicine ), with 78% of their articles made OA within 12 months, and 85% within 24 months. However, these findings may not apply to the field of dentistry. To our knowledge, among the current top-20 high-impact dental journals only one ( Operative Dentistry ) adopts a Delayed OA model, with an embargo period of 36 months .

3.3

Gold vs. Green

There has been a debate over which of the two strategies is more cost-effective . Some claim that Gold OA is the only sustainable model for future publishing, and that governments should allocate funding to promote Gold OA publishing during the current transition period (from fully subscription-based scientific communication to 100% Gold OA) . Others argue that Green OA is more feasible and reasonable, and that adopting only the Green strategy during the transition will not incur sustainability issues in the future .

The advantages of Gold OA over self-archiving mainly include:

• •

  Availability upon publication (without any embargo, in the cases of “Direct OA” and “Hybrid OA”);

• •

  Removal of permission barriers (with licenses allowing re-use, i.e. “libre OA”);

• •

  More reliable content and presentation (especially when compared to author’s preprints) ;

• •

  Ensured persistence (with no or low risk of missing documents); and

• •

  No space or length limits for the article (unlike traditional subscription-based printed journals).

Green OA, on the other hand, has the following advantages over Gold:

• •

  Lower costs, at least from the perspective of authors and institutions (the average cost of setting up and maintaining repositories is estimated at between 2 and 53 USD per article, much lower than the average APC of 1100 USD) ;

• •

  Lower risk of incurring deceptive or irresponsible publishing practices (see Section 8 ) ; and

• •

  Lower risk of making the available evidence pool biased towards funded studies, which are more likely to be published in OAJs and may have different findings from unfunded studies .

4.1

Development of the internet

Although some have claimed that the OA movement can be traced back to the 1960s , modern OA publishing only happened after the advent of the Internet and World Wide Web in the early 1990s . The emergence of electronic journal collections and online bibliographic databases changed the way people use scholarly publications, making them less reliant on printed journals and personal subscriptions . The Internet also made it possible to substantially decrease the costs of scientific publishing, as those costs for printing and shipping (estimated at 30%–80% of the total budgets of print-based journals) were no longer necessary in an online-only publishing model . In addition, online open repositories for the self-archiving of scientific articles began to emerge. In 1991, the American physicist Paul Ginsparg founded the first of such repositories−arXiv.org, which soon became the primary means of communication for physicists . PubMed Central (PMC), the first and most successful subject-based open repository for biomedical journal literature, was launched by the U.S. National Library of Medicine in 2000 .

4.2

The “serials crisis”

Another major motivating factor for OA is the financial crisis faced by research libraries, caused by rapidly rising subscription prices for scholarly journals and the relatively static budgets of libraries. During 1986–2005, for those member libraries of the Association of Research Libraries (ARL), their overall subscription expenditures tripled, but the number of purchased items only increased on average 1.9% per year . Most libraries had to pay more but provide their users with less.

One main reason for the “serials crisis” was the oligopolistic structure of the journal publishing industry, which had resulted in negligible rivalry among major commercial publishers and their high profit margins. Around the year of 2000, it was estimated that Elsevier, Springer and Wiley published about 42% of all journal articles; and that the average operating profit margin of Elsevier was as high as 36.4% for their journals in science and medicine .

In light of the above, libraries and librarians became the principle advocates of OA. Their efforts have led to the emergence of many major OA organisations [e.g. SHERPA ( www.sherpa.ac.uk ) and SPARC (Scholarly Publishing and Academic Resources Coalition; www.sparcopen.org )], OA initiatives/statements (e.g. the Bethesda Statement ), institutional repositories , as well as numerous research studies about OA .

4.3

Ethical and equality issues

In the traditional pay-to-access publishing model, access to the scientific literature is restricted to those researchers and professionals in wealthy institutions. Coupled with the issue of “serials crisis”, subscription (and therefore access) to all journals has become unaffordable to any single individual, library, university or research institution . Such restricted access has lowered the value of research and led to ethical problems.

Firstly, most research projects are funded by citizens through their tax. It is unreasonable that citizens have to pay again to access the research output. Secondly, access to knowledge should be regarded as a human right . Restricted access to scientific knowledge favouring higher income countries is harmful to global human equality and global public health . Thirdly, all peer-reviewed articles are written for the purpose of being used and being built upon by all potential users, not for royalty income of the authors or publishers . In this sense, costly subscription fees and limited dissemination of these articles are also against the intentions and efforts of scientists and scholars.

In January 2002, the World Health Organisation (WHO) and six major publishers started the HINARI (Health InterNetwork Access to Research Initiative; www.healthinternetwork.org ) programme, which provides local, non-profit institutions in developing countries with free or very low-cost online access to major biomedical journals. In August 2004, the Alliance for Taxpayer Access (ATA; www.taxpayeraccess.org ), a coalition of patient groups, physicians, researchers, educational institutions, publishers, and health promotion organisations, was launched in the US to advocate barrier-free access to taxpayer-funded research .

5.1

Immediate dissemination

When an article is published in an OAJ, the final edited content of it after peer-review is available online upon publication. In the case of a preprint, self-archiving upon (or even earlier than) acceptance for publication is allowed by most journals, which means that delay in dissemination due to the editorial process is also avoided. Such early availability and dissemination may improve the time relevance of articles, accelerate the progress of research, and benefit all direct and indirect users of research (e.g. patients) .

5.2

Good findability

As the metadata of OA articles (usually the full-texts, not just abstracts) are “exposed” on the World Wide Web, they should be easier to identify than non-OA articles, especially when using general search engines like Google . In addition, Green OA copies archived in those repositories that meet interoperability standards (e.g. the OAI protocol) can be easily identified using the search functions of some OA specific databases, such as the OAIster and Open DOAR.

5.3

More downloads

Several studies have shown that OA articles are downloaded more, and by more people, than non-OA articles. In a randomised controlled trial (RCT) using journals in physiology, Davis et al. found that articles assigned to the OA group received 89% more full-text downloads, and had 23% more unique full-text visitors than those assigned to the subscription access group. More recently, based on a study of papers published in Nature Communications , Wang et al. found that OA articles had a greater number of total downloads, that were sustained and consistent for a much longer time than non-OA articles.

5.4

Broader societal impact

The emergence and development of “altmetrics” (alternative metrics–a range of measures of research impact that go beyond citations) since 2010 has provided a novel method to measure the societal impact of research articles . Major altmetrics services, such as the Altmetric.com and ImpactStory, have been tracking and analysing the digital use/sharing of articles, based on data collected from policy documents, news outlets, blogs, social media (e.g. Facebook, Twitter), Wikipedia, online reference managers and other sources . Wang et al. compared the Altmetric score of OA and non-OA articles, and found that OA articles received 20%–48% more discussions in social media than non-OA articles.

In addition, increased access to the scholarly literature is beneficial for the fostering of a culture of greater scientific education and literacy, and the improvement of public policies and public engagement in scientific research .

5.5

Reducing research waste

In 2009, Chalmers and Glasziou estimated that three problems (flawed design, non-publication, and poor reporting) had led to an avoidable waste of 85% of all research funds in biomedicine, indicating a total global loss of more than 100 billion USD per year. However, when the published report of a research study is not effectively communicated to those who could benefit from it, this research, no matter how well designed and reported, is at least partially wasted. In the case of dental and medical research, the main beneficiaries include not only researchers, but also clinicians, patients and policy-makers, who usually have limited access to full articles behind the paywall . Therefore, further development of OA can help provide a comprehensive, accessible and unbiased evidence pool to all stakeholders, and thereby reduce waste in dental and medical research .