Recommended: a remarkable and very complete review of OA by Peter Suber.

The “Green Open Access (OA)” solution, providing free access to research publications in Institutional Repositories (IRs) via the Web, is certainly the best one, but sooner or later it will face a new wave of centralised thematic or funder repositories (CRs).

The latest initiative comes from the very active EUROHORCs (European Association of Heads of Research Funding Organisations and Research Performing Organisations), well known for its EURYI prizes and for its prominent influence on European thinking in the research area. EUROHORCs is working to convince the European Science Foundation (ESF) to set up, through a large subsidy from the EC, a centralised repository (CR) which would be both thematic (Biomedical) and local (European). The concept is inspired by PubMed Central, among others.

The EUROHORCs initiative is very well-intentioned. It is based on an awareness that many of us share: It is of the utmost importance that science funded by public money should be made freely and easily accessible to the public (OA). But the initiative also reveals a profound misunderstanding about what OA and researchers’ real needs are all about.

The vision underlying the EUROHORCs initiative is that research results should be deposited directly in a CR. However, if research results are not OA today, this is not because of the lack of a CR to deposit them in, but rather because most authors are simply not yet depositing their articles at all, not even in an IR.

Creating a new repository is hence not the solution for making research OA. The solution lies in universal deposit mandates, from both institutions and funding agencies. If this task is left to large funders such as the European Community, their central repositories will only contain publications of the research they have funded. From this it is easy to see that researchers will ultimately have to deposit their publications in as many repositories as there are funders supporting their research. Not only is this not practical, it is needlessly cumbersome.

The obvious solution is that both research institutions and funding agencies should jointly require IR deposit. Once that systematic coordination has been successfully implemented, if CRs are desired, they can easily be created and filled using compatible software for exporting or harvesting automatically from IRs to CRs.

What is worrisome is the needless double investment in creating two distinct kinds of repositories for direct deposit. This trend seems to rest on the naive notion that, in the Internet era, it is somehow still necessary to deposit things centrally. But in reality, the centralising tool is the harvester, and its search engine. Google Scholar, for example, is quite efficient in finding articles in any repository, institutional or central, yet no one deposits articles directly in Google Scholar. The perceived need for direct-deposit CRs is groundless, technically speaking. Such CRs even run the risk of serving as hosts for only the publications funded by a single funder. IRs guarantee OA webwide for all research output, in all disciplines, from all institutions, regardless of where (or whether) it has been funded.

It is understandable that funders may wish to host a complete collection of the research they have funded, but nowadays that can easily be accomplished by importing it automatically from the more complete collections of the distributed IRs — since institutions are the universal providers of all research output, funded and unfunded — as long as funders collaborate with institutions in first ensuring that all the IRs are filled with their own institutional research output.

Besides, the OA philosophy is global. It cannot be reduced to a single continent. Science is universal.

Giving priority to creating more CRs for direct deposit today is not only a waste of time: it is also counterproductive for the growth of convergent funder and institutional mandates. It would generate multiple competing loci of primary deposit for authors — most of whom, we must not forget, are still not depositing at all.

In conclusion, it seems far more efficient to focus first on filling IRs at this time; once that is accomplished, if it is judged useful, CRs can be configured to collect their data from IRs rather than being used as divergent points of direct deposits themselves.

The potential success of OA, without conflicting head-on with publishers, rests on the deposit of authors’ own final drafts of their published articles, through a one-time, simple action on the part of the author. All research is generated from research institutions: IRs are hence the natural locus for author deposit, providing optimal proximity, convenience and congruence with the mission of the author’s own institution. The rest is merely technical: a matter of automated data transfer to external CRs.

The EUROHORCs proposal is only worthwhile if it contributes to the secondary harvesting of data from primary IRs. Otherwise, it is missing the point of OA.

ORBi wins its challenge

U. Liège’s IR « ORBi » (Open Repository and Bibliography) is fulfilling its promise: over 4,000 references have already been filed since November 26th and, in a happy surprise, 79% of these articles turn out to be full text. This is thus ahead of schedule for our institutional Green OA Mandate (announced in March 2007 to take effect in October 2009): « Whenever the university reviews faculty publications for promotion, tenure, funding, or any other internal purpose, the review will be based exclusively on full texts deposited in the IR. »

The graph below shows clearly how the IR contents are growing. And yet a quick calculation also reminds us that we are still far from capturing the actual number of papers published yearly by our university authors.