Contents

  Freedom of Information Conference 2000

Harold Varmus
Memorial Sloan-Kettering Cancer Center


What will the new environment look like?

Most of us are here today because we recognise that science and the publishing of science in particular is undergoing a fundamental but also difficult transition. A transition from a mode of publication that arguably began 500 years ago with the publication of the first scientific journals and a transition to a new mode that's made possible by computer science and the internet. We all know and believe this transition is going to occur and I suspect we're all in this room today because we believe it's going to occur in the next decade or two. But all scientists and publishers of science are going to have a strong influence over the pace and form this transition takes, and I think that's the charge for today and for tomorrow; trying to figure out how this transition occurs and how quickly it occurs. So, in addition I think today is a time to consider where we are in the transition, how the early steps have occurred and to think of corrective actions-just as if we were sailing toward a distant port and had to attack and win.

Five goals for electronic publishing
The first goal seems to be to broaden and simplify access to scientific data. This, of course, includes better dissemination and more thorough presentation in the view of many of us. Access itself should be completely free and that, of course, is an arguable point I'm sure will be discussed later. But in any case, from the perspective of someone like me who's a biomedical scientist, the point of increasing access is to accelerate the use of scientific data for the benefit of public health.

The second goal is economy and it's the conviction of many of us that in the long run the use of electronic publishing will reduce costs of publishing overall, although the distribution of expenses may be argued - that is, how various components of scientific publishing share those costs, such as authors, readers, publishers, scientific societies, institutions like the House Scientist, Government agencies and libraries, and the degree to which those components should share the costs is a debateable issue. But I think it's fair to say that in the long run economies can be achieved.

The third goal is quality. It seems to me that one of the possibilities electronic publishing offers is to preserve and even improve the quality of the review and comment about manuscripts. This is an issue that has been hotly discussed because of the possibility of contaminating the worlds of literature through random and unfiltered access to the distribution source. But it's my conviction that we can sustain and create peer review journals and that we can find new ways to critique and supplement papers in a way that will actually improve quality through electronic means of distribution.

The fourth goal no one argues with is innovation. There are ways through digital presentation to improve the way we portray our results by making large searchable data sets embedded in scientific manuscripts, by providing pre-print servers, by using hypertexts, by using abundant digitalised images and even movies, all of which can markedly enrich the reading experience.

And the final point, of course, is retrieval. This is something we already experience and benefit from the idea that we can archive, search, and retrieve reports in powerful ways, placing large amounts of the scientific literature, including past literature and, hopefully, present and future literature into data bases of multiple sites with innovative search engines that allow scientists to carry out that second reading function - the first being browsing, the second being searching. The second probably for most part the more important.

The issues we overlook
There has been resistance towards proposals to how we should sail towards these goals and, as a highly visible author of manifestos for systems such as E-BioMed, E-BioScience, and now PubMed Central, I've been one of those people who's felt the resistance strongly. Now, before proposing how the ship should sail in the short run I'd like to talk about three fundamental issues that are often overlooked in the kinds of discussions we're having today.

The first issue, for brevity, I refer to as "writers versus authors." The point here is to emphasise that a person describing scientific results is motivated in a fundamentally different way from people writing other kinds of prose. I think about this by thinking about my personal computer. My wife and I spend a lot of time composing sentences on that computer and thus we are both, in some sense, writers. But that's really where the similarity ends. When she writes about philanthropy or about gardening she hopes to sell her material. The texts for these ideas, its news, its sentences are sold to a magazine, newspaper or book publisher who will then sell them to interested readers. And she wants lots of readers because every time a new reader is added she gets a larger payment. She has very little interest in distributing her work for free because writing is what she gets paid to do. She is, what we call, a "writer."

When I write, especially when I write a scientific report, I feel more like someone trying to disseminate results generated in the laboratory rather than a real writer. I'm hoping my report will be favourably viewed by an editorial board of a well-read and respected journal not because I want to collect a fee but because that's how my work can best have an impact on other scientists and best have an impact on the development of the field. After all that's why the NIH and the American Cancer Society have paid for the salaries, equipment, and supplies that allow me to do the research. So, as an author of scientific reports, what I'm interested in is having my work viewed by the greatest number of people who are likely to have some interest in it. This is more likely to happen if it appears in a journal that many people read because they respect the journal but they're also especially likely to read it if they have access to it. None of us can directly control the reputations of journals but we can improve access to those journals. Expanded access then has a tremendous advantage to authors and readers and the public that benefits from science.

The second issue concerns the disenfranchised, a topic we're all aware of but not spoken about enough. The effect of electronic distribution of scientific information on the disenfranchised can be even more profound than some of us think. In technologically advanced and wealthy countries such as ours, scientific reports are not hard to get hold of. We have good libraries, we can afford to subscribe, we use PubMed regularly, and often our institutions subscribe to the technically excellent vendors like High Wire that make an important selection of journals accessible to our institutional computers. While these mechanisms may be imperfect they provide a good service and frankly more reading material than most of us have time to deal with. Outside of this narrow circle of affluence, however, access drops precipitously. Even in the US and Europe many researcher institutions don't have subscriptions to electronic texts or the full panoply of journals and most investigators can't afford more than a few journals of their own. In the developing countries the situation is dramatically worse and it seems especially unfortunate when we recognise that putting scientific reports on-line for the rich countries would tremendously benefit those in poor countries that really have acute problems of access.

I recognise this is not simply achieved by putting journals on-line for free. There is a need to establish strong internet connections and to put technical service on the ground. My only experience in overseeing research on malaria in Africa has taught me this lesson. We had to send people from the National Library of Medicine out to a variety of countries to try to establish even a single sight of internet connectivity. In fact, the importance of electronic publishing on the disenfranchised has not been overlooked by Kofi Anan, the Secretary General of the UN. In his millennial statement issued a few months ago he talks about digital bridges and, in particular, on the impact of electronic distribution of information in the health and science arena. He says, for example, that new technology offers an unprecedented chance for developing countries to leapfrog earlier steps of development and everything must be done to maximise their people's access. He points out that information has unique attributes that have tremendous potential benefit to some of these countries - steel, boots, and other things are consumed, information is different, it can be made available for multiple use and is not consumed. In fact, it's more valuable the more it's used. He urges the policy-making world to understand how the economy of information differs from the economy of scarce physical goods and to use it to advance policy goals such as a new health inter-network for developing countries to establish 10,000 on-line sites and to transmit health and medical information tailored for specific countries. He also announces a second digital bridges initiative called the United Nations Information Technology Service that will be used to train groups in developing countries to be the technical forces on the ground, to ensure that the installation of these new tools will bring information in a sustained way to these developing countries.

The third issue concerns the filtering of information on the internet and the retention of useful hierarchies. It's my conviction that distribution of scientific information through electronic means does not mean having unregulated or unqualified dissemination of information or a loss of a valued hierarchy of information structure in science. In the responses that David Lipman, Pat Brown, and I received to some of the proposals we put together last year, the greatest passion was invested in this notion. I understand why it's easy to be frightened about what's happening, through the internet, in many disciplines and could happen in science. We shouldn't be naïve about this.

This weekend I was reading about a different politics and I was struck by a paragraph on an essay by Dick Morris in a book about recent political tracks. The writer said, and I quote, "Morris argues that modern technology has made voters better informed than ever and thus better qualified to take a more direct role in law making. I doubt this. There is certainly more information available to more people but the internet has also removed the traditional filters that once screened a good deal of nonsense out of our national debate. On the web today one can read all manner of conspiracy theories, baseless accusations, character assassinations, and economic quackery. Many people think no doubt, this stuff could not be said if not true, in some place there was an authority who would keep the record straight. There is none." Now, obviously this concern - which is a real one - pervades much of the discussion on scientific publishing. We all know there's a tremendous amount of junk out there so there is no doubt there is a problem. But I believe it can be controlled if we are careful and there are many ways to do that; one is by credentialing contributors to any electronic sites that are established for scientific publishing, labelling entries with the explicit criteria that it used for exclusion, retaining many of the filtering properties that currently exist in print journals, editorial boards, peer review and many others.

Achieving the transition
I'd like to talk the last few minutes about what I referred to at the outset as the transition, that is this potentially short, potentially prolonged period in which we go from a world of paper-based journals to an electronic world. One of the difficulties about making proclamations as the ones that my colleagues and I made last year is that one tends to get rather specific about how the final world should look and I think it's premature to say that. Instead I'd like to think about what we ought to achieve in the next couple of years to bring us closer to where I think we all want to be.

First is testing the idea of open access, to demonstrate what it feels like to have an electronic distribution of a few important journals. I would contend that the activities currently in progress with PubMed Central are an example of the way in which we contest what open access is like.

The second objective is data collection, developing data that establishes the technical and financial boundaries of unfettered electronic distribution. Many of the debates I've engaged in have been characterised by a paucity of information about what is feasible and what the real costs are. This will allow more solid proposals for operation of large systems and I would again argue that many of these data collection exercises could be carried out in the context of the PubMed Central experiment.

The third objective is the initiation of some new electronic journals specifically targeted for distribution by a mechanism like PubMed Central. It's no secret that BioMed Central represents a collection of such journals and I congratulate Vitek Tracj for making this leap into the unknown.

The fourth issue is studying the consequences of starting such journals. How are they received? What are the impacts on library costs? What are the attitudes toward electronic journals published only electronically with free access for the reader? What is the willingness of authors to publish in such journals as opposed to paper journals of more limited circulation?

The fifth issue is one I think is particularly important and that is the electronic world has the potential for storing information published from the past that has become more and more inaccessible with time. One of the greatest things we can do is to build a large electronic archive of the important literature in biomedical, chemical and allied research and that we can do it without extraordinary expenditure - one that will definitely pay off in the long run. But we need to begin to experiment with this, determine how much it's going to cost and begin to line up potential financiers to make it happen.

The sixth issue is the creation of mirror sites. Ever since our first manifesto we've been talking about having multiple sites and an international network of distribution sources and archiving sites. It hasn't happened yet and I think it's very important that our colleagues in Europe and Asia begin to consider teaming up with PubMed Central to develop mirror sites so we have assurance about storage and distribution and that we emphasize the international aspect of scientific publishing.

The seventh issue is one that is close to my heart but a controversial one, and that is that we begin to experiment with a more porous filtering process for scientific data that's presented through such means as PubMed Central. We should do some experiments with screened but un-reviewed reports, perhaps starting with things most highly acceptable for such distribution like sequence comparisons, protein structures, big data sets, resource inventories, and clearly indicate the material has not been through the traditional peer review process. I firmly believe doing experiments of this kind will present to the reading public some remarkable ways in which innovative thinking about electronic publishing can markedly enrich the experience of doing science in the modern age.

And of course the final issue is a platitude but it's an important one, and that is that we keep talking, we have meetings like this one in which people representing the concerned constituencies keep the issues in the air by writing opinion pieces, seeking support from groups outside the usual sets of people including advocates for disease specific research, health workers, politicians, and many others.

Questions from the floor

What exactly is PubMed Central?

Questioner 1: My name's Bob Simone from Stanford University (associate editor of the Journal of Biological Chemistry). What exactly is PubMed Central? I don't ask the question lightly. I've followed its development quite closely. It was going to be a pre-print server, it was to be an archive, it was to be a journal, and watching what's transpired since it started six months ago, it's not obvious that it is any of these. It's also not obvious where it's headed.

Harold Varmus: First of all, I don't agree with your characterisation of what it was intended to be before. No one ever argued this was going to be a journal but I do agree with you that it has undergone changes, although the basic vision has not changed.

The first vision that was laid out was the vision of one set of blueprints but maybe not the right set of blueprints and I think what we've progressed to is a short term view of how a government agency can help us move toward a world in which the long term goals can be achieved. What the short-term view represents is a distribution site, in Pat's metaphor, a post office for distribution, a public vehicle with government financing to help distribute electronic information published in existing journals.

When would they be presented to the public through these means? It could be at the time of acceptance. It could be at the time of publication. It could be a month, two months thereafter. Most journals that have contributed to PubMed Central have elected to do so roughly a month after publication, obviously designed to protect their subscriber base. One of the glorious things about the way it's been set up is that it's intimately connected to PubMed itself, which is a widely used search engine for titles and sometimes for abstracts. So you go to PubMed, get your title and if that journal happens to have provided its content to PubMed Central you have immediate access to it. I'm not sure how many journals currently provide material but it's something in the order of 20.

Questioner 1: So it is currently a repository?

Harold Varmus: Well, it's also a distribution mechanism.

Questioner 1: What about the other components? I mean, there was a time when there was to be a peer review component.

Harold Varmus: There was to be. It was going to encourage the formation of journals, a screening process that allowed people to post things but would not be a peer review system run by the government.

Questioner 1: So it's currently a repository in the sense that it's posting and distribution. What about the pre-print server, for example?

Harold Varmus: Those are things that could happen in the future. There is an advisory board which I and several people in this room serve that has discussed the issue - whether to establish a pre-print server and take on non-reviewed material that is screened but not reviewed. We decided to defer that decision to a later time.

Will PubMed Central result in the end of scientific societies?

Questioner 2: The internet offers some wonderful opportunities for scientific societies and also risks if they make a false step. For many of these societies the publication of a journal is their primary source of income. If these societies lose that income it will wipe out their activities. What will happen to science if these societies are not able to exist?

Harold Varmus: Putting the question that way is very different from asking how the societies can contribute to a beneficial development to do with publishing. Science could not thrive without scientific societies, they're incredibly important, they are our gills and no one interested in scientific publishing thinks they should disappear. However, I do think societies should not be reacting reflexively and defensively to proposals that would benefit them by improving the environment in which scientific publishing is done because they think they're going to lose revenue. Instead, other kinds of solutions should be sought and I believe there are other solutions. For example, it seems to me that most societies can generate revenues in other ways and to say the existence of the society depends exclusively on the generation of profits from publishing is to my mind a very unfortunate thing to have to say, I don't believe it's true.

There are societies that do well without making profits from journals. I think this may require multiple years of changing the mind set, especially of young scientists, towards the function of scientific societies. These societies are our unions. They work for us, they do many things, and they don't simply exist to publish journals. They exist to nurture the careers of young scientists, to help in the public debates about the importance of science, and to help the government see the wisdom of supporting science through government agencies. All these activities are important and may require raising dues. I can't imagine people failing to support their society because the society has helped to advance innovations and publishing to make science a richer experience but it will take time to change attitudes.

I know there is a time when you are encouraged to join societies because you're told that if you join a society you will get some journals at a cut rate. But a change in society policy that promotes the completely unfettered access towards all journals should be something societies agree to. I think we may have to encourage a sense of citizenship and public responsibility toward the support of scientific societies. I actually believe that is realistic and can happen, assuming it is done in a measured way.