The internet has sparked a revolution in scholarly communication, the process by which researchers share new ideas and findings. For universities, this raises a number of technical, managerial and financial issues. The experiences of the Open University are a typical example.
Widely recognised as a distance-learning institution that awards undergraduate degrees, the university currently has about 218,000 students around the globe. What is less well known is that it is also an active research centre, with more than 1,000 research staff and 500 postgraduate students.
Like their peers in other universities, when Open University researchers want to disseminate their ideas more widely than the lecture room, they turn to scholarly publishers - learned societies and commercial publishers such as Wiley, Elsevier and Springer.
These publishers arrange for the research to be quality checked - by means of peer review - and then distribute it in journal or book form. Each year, Open University researchers produce more than 1,000 books, book chapters and peer-reviewed journal articles.
Like other researchers, the Open University faculty also turns to scholarly publishers when it needs access to externally produced research. To that end, the university library buys many new books each year, and subscribes to thousands of scholarly journals, most of which are now distributed online.
In fact, scholarly publishers have been key intermediaries in the scholarly communication process since at least 1665, when Henry Oldenburg created the first peer-reviewed journal.
But in recent decades, the quantity of research produced has grown so rapidly that the system has begun to creak, particularly in relation to journals. The content of every journal is unique, so each one is, in effect, a monopoly. And because researchers expect to have access to every paper published in their field, publishers can set high prices and make large profits. The result is that universities are struggling to fund their journal habit.
Over the last seven years, for instance, the Open University has seen its electronic journal subscriptions bill more than triple, from £284,000 to £997,000. Add to this its remaining print subscriptions and the university's annual journal bill is nearly £2 million. Even so, it cannot afford all the journals it needs.
From this crisis, the open access movement developed. The fundamental problem with the traditional publishing model, say open access advocates, is that it requires the research community to give publishers its research for free, undertake peer review for free, and then buy back that research in the form of increasingly expensive journal subscriptions. As most of the costs of distribution disappear on the internet, costs should be falling, not rising, they argue.
And because research published in scholarly papers is funded by the public purse, it should be free to all, says the open access lobby. But publishers demand, as a condition of publication, that authors assign all rights to them - so it is effectively privatised. The public is then asked to pay again for researchers to read it.
The solution, open access supporters suggest, is for researchers to self-archive their papers on the web, liberating themselves from the subscription firewall.
Self-archiving does not mean researchers will stop using publishers. Rather, they make electronic copies of their postprints available as a free supplement to the publisher's version. With this aim in mind, many universities have created their own web-based institutional repository as a central location for researchers to deposit their papers.
The Open University's repository was created by its science librarian in 2002, and is based on an open source software package called GNU EPrints. The initial aim, says Open University research support librarian Bill Mortimer, was to allow anyone in the world to access the Open University's research, regardless of whether their institution could afford the publisher's version. This also enabled the university to show that it is not just a degree factory.
But it soon became apparent that self-archiving features pretty low on the average researcher's list of priorities. And running a repository is not costless. "Although EPrints is open-source software, there are a lot of support costs," says Mortimer. "So the science librarian was not able to support the repository on an ongoing basis, and eventually it was shelved with about half a dozen items in it."
However, the repository was reprieved in 2005, when Mortimer was appointed. Tasked with supporting Open University research staff, he launched an advocacy programme, and began attending faculty meetings. His pitch was: open access fits the ethos of the Open University like a glove, and benefits the university faculty as well as external researchers.
The Open University was founded in 1969 by the then Prime Minister, Harold Wilson, and its mission is to be "open to people, places, methods and ideas", says Mortimer. What better demonstration is there of that than to support open access?
Mortimer also drew the faculty's attention to studies showing that making scholarly work freely available on the web significantly increases the number of times the work is cited. Citations are highly valued by researchers, not just for the kudos they give, but because they increase the chances of being promoted.
But it was a slow, laborious process, says Mortimer. "The majority of Open University researchers are, theoretically, very positive about the idea of providing resources to people who don't have membership of a library, and there were always a few people who became very enthusiastic after I talked to them. But it was hard to move beyond a few core enthusiasts."
This is a common experience. Studies have shown that only 15% of researchers spontaneously self-archive their research outputs, even when encouraged to do so.
But a turning point came in 2005, when the Open University appointed Brigid Heywood as pro-vice-chancellor research. With the national Research Assessment Exercise (RAE) fast approaching, Heywood needed a centralised record of the research output of every full-time member of the university.
Undertaken every five years to evaluate the quality of research done by Britain's higher-education institutions, the RAE is important because a university's funding is based on its RAE score, which is calculated by assessing the quality of four research outputs for every member of staff.
But the information Heywood needed was spread across a patchwork of departmental systems, most of which were incompatible, and many of which had incomplete records. However, the Research School discovered that the library had the foundations of something they could use for the RAE, says Mortimer. "So, in early 2006, the institutional repository underwent a major reincarnation and was relaunched as Open Research Online, or ORO."
In the process, the EPrints software was upgraded and an RAE plug-in added. This enabled reports generated from ORO to be imported directly into the system used by the Higher Education Funding Council (HEFCE), the university funding body.
ORO's adoption by the Research School provided vital funding. It also enabled Mortimer to significantly increase the number of records it contained, and between January and July 2006 alone, he harvested between 2,000 and 3,000 items from departmental databases. Today ORO has more than 6,000 records, including peer-reviewed journal articles, books and book chapters, conference papers and patents.
But the downside of Research School support for ORO was that the original objective of making the Open University's research freely available was diluted - for RAE purposes, it was enough to input the bibliographic details of researchers' publications. "It has had to be a kind of balancing act," concedes Mortimer. "I guess the RAE has put the weight on one side of the scales."
Nevertheless, open access remains an important goal for the Open University, says Heywood. While stressing the need for ORO to "provide management information about research activity, provide support to researchers, and profile the expertise and richness of the university's research portfolio", she also insists it is "critical that the outputs from publicly funded research are disseminated and shared with the widest possible audience".
Publishers dislike self-archiving, but faced with growing restiveness over journal price inflation and simmering anger at their profits levels, many have come to accept the practice - although often insisting that self-archived copies are made available only after a six- or 12-month embargo. And 38% still do not permit it.
However, as the Open University has discovered, the greatest obstacle to filling institutional repositories remains researcher recalcitrance. Even now, only 15% of the records in ORO are full-text.
Research funders have therefore begun to make self-archiving compulsory. In October 2005, for instance, the UK-based Wellcome Trust introduced a mandate requiring all papers resulting from research it has funded to be made freely available on the internet within six months of publication.
Most UK research councils have followed suit, and last December the world's largest funder of medical research, the US National Institutes of Health (NIH), also introduced a mandate. NIH-funded researchers must ensure their papers are freely available within 12 months of publication.
There are now 22 other funder mandates in place, plus 12 institutional mandates, including one at CERN, the world's largest particle physics lab. Four universities departments have also made self-archiving compulsory, including a forthcoming mandate at Harvard's Faculty of Arts & Sciences.
How effective mandates will prove remains unclear. A recent survey by the Wellcome Trust, for instance, found that only 27% of the papers it had funded were freely available six months after publication.
However, publishers are conscious that self-archiving poses a long-term threat to their profits, and have begun to adapt. On payment of an "article processing charge", most will now make papers freely available on the web themselves - and from the day of publication.
Pioneered by next-generation publishers such as Biomed Central and Public Library of Science, this "open-access publishing" model reverses the traditional arrangement whereby readers pay to read papers, to one where authors - or, more usually, their funders - pay a one-off, up-front publishing fee.
The key question for universities is whether open-access publishing will take costs out of the system, or simply transfer them from the library to another part of the institution. There have already been a number of controversial article processing charge price hikes, and rates currently range from about £250 to £2,400 per article.
But publishers surely realise that if they want to retain the intermediary role they have played for the past 350 years, they will need to control their prices. After all, in the age of the internet, universities could do it all themselves.
For the Open University, the issue is whether to introduce its own mandate. "The university is currently reviewing mandatory engagement with ORO and is taking advice and guidance from other institutions and agencies that have developed such policies," says Heywood. "We expect to reach a decision by the summer of 2008."