Publisher Perish.

The coming battle over putting all scientific journals online for free.

RICHARD ROBERTS, A NOBEL LAUREATE for his work on nucleic acids, has become a major pain-in-the-butt for the scientific publishing industry. Like many highly successful scientists, Roberts has a reputation for being thoughtful, careful, and thorough. But he's now working on an idea that could turn scientific publishing on its head.

Along with a large group of scientists--including Harold Varmus, the acclaimed former director of the National Institutes of Health (NIH)--Roberts has been pushing the buildup of a massive repository of scientific literature called PubMed Central, hosted by NIH's National Library of Medicine. According to its advocates' vision, PMC would allow anyone in the world to freely search and retrieve the full text of any published scientific article, ideally with archives going back for decades. Want to know what the Journal of Cell Biology and all of its peers had to say about protein extraction last year, or in 1972? Log in; enter search terms; presto.

The way the system works now, anyone can log into a database hosted at NIH known as Medline, type in search terms, and retrieve article abstracts--a service currently used several hundred million times per year. But anyone who finds a useful reference has to trek over to the library, photocopy the article from the bound journal, and pore over it for the specific references. If there's no library around, too bad. NIH has a version of PubMed Central up and running, but with only about 50 journals on board, few people use it.

PubMed Central and Medline may not seem that different. But scientists and publishers alike agree that it would be revolutionary to pass from searchable abstract to searchable texts. People in the Third World would suddenly have access to the planet's great libraries; lay people interested in specific diseases would have the best information at their fingertips; all current medical researchers would save countless hours and could investigate their work much more thoroughly. An abstract can tell you the basic outline of an article--but it doesn't say what's in the data, what's buried deep, and what might have been discovered on the side, all potentially the difference between life and death. According to Varmus: "Think about hexamethonium, the drug that killed a student at Johns Hopkins. There's research out there that showed its danger. With PMC, people might have known that."

Or consider the case of Jens Mielke, a researcher at the medical school of the University of Zimbabwe in Harare, the country's leading health center. Mielke is part of a team desperately trying to diagnose and help treat cryptococcal meningitis, a form of the brain and spinal disease which afflicts HIV-positive men and women and is usually lethal if left untreated. The hospital treats about 500 meningitis-sufferers a year. But the medical school only has valid and up-to-date subscriptions to about 20 journals, compared to the thousands available in large American facilities. Consequently, Mielke lacks access to, or knowledge of, most of the current diagnostic work in his field. If his team could access PubMed Central, he says, it would "make an enormous difference. It would enable us to diagnose at twice the efficiency."

Death Be Not Proud

PMC would also have a revolutionary impact on the scientific publishing industry, and the thousands of journals catering to the medical-research community. According to Karen Hunter of Elsevier Science, a giant publishing company that puts out about 1,100 journals ranging up to about $17,000 for annual subscriptions: "The number of subscription cancellations would be significant. A lot of magazines would close."

That doesn't sway everyone. A group of scientists, including Varmus and Roberts, have organized into an entity called the Public Library of Science (PLOS), putting out a pledge not to publish in, subscribe to, or peer review for, any journals that don't provide open access through a large central database, which in practice means journals that don't give their information to PubMed Central. About 27,000 other scientists joined them, though the group backed off a bit when the actual boycott date arrived on September 1st. Now, instead of a strict boycott, the group plans to start new journals, and give priority when trying to place their articles to those complying with PubMed Central. Despite the minor retreat, the group still has attitude. "Traditional publishers," Roberts says, "are useful and there's a role for them. What's obsolete is the way they're doing business."

That may be an overstatement, but Roberts has a vital point. The government funds scientific research across the country for the public good, and for years the system that we have today--government pays for research; private journals publish research--seemed the best alternative. Now, the Internet has changed the nature of the game. Information is much more mobile and, therefore, there's an increased value to making as much as possible public in an unbiased way--a role suited to government. The potential gains for medicine are extraordinary, even if a few people may end up losing their shirts.

Club Med 

The Internet was supposed to change all forms of publishing. E-books were the future and print magazines and newspapers strove to Get Big Fast online. The New York Times' dot-com division talked about a massive IPO, and Salon threw the hippest parties in town. Both may again prosper, but for now they're holding out their hats. One year ago, Inside.com, then a sparkling new online news source, published an article panning a dinosaur called The Washington Monthly. Today, Inside has been gutted and swallowed up by a competitor.

But scientific publishing is different. Dot-com journalism sank because no one has yet figured out a viable business model that allows an online publisher to pay salaries, hire freelancers and editors, and support Web programmers. But scientific journalism shouldn't have those worries. The people who write for scientific journals do it for prestige, for their colleagues working on similar projects, and perhaps to earn promotions; they don't do it to put food on the table since other people pay their salaries. Peer reviewers of scientific papers usually work for free as well; they just want to help the cause. Perhaps most important of all, the entire industry relies on handouts from the government. The people who write for The Journal of Cell Biology work on projects supported by the government in federally funded laboratories. The Public Library of Science just wants the government to carry that research one step further, disseminating it as widely as it can.

To Roberts, Varmus, and the other scientist advocates who comprise the PLOS, the new model for scientific publishing should work like this: Researchers would prepare their papers and submit them to journals, paying the journals for the privilege of being published. The journals would then subject the articles to current standards of peer review--sending each paper out to several knowledgeable people in the field--and then publish the accepted articles. The journals would then send out their bound issues as they do today. Six months later, they would file with PubMed Central. On its Web site, the Public Library of Science writes that publishers would be able to make money "in the same way that a midwife can earn a living without keeping the babies that ,he delivers" PLOS would also create its own set of journals that would file directly with PubMed Central and that might not even have print versions, charging authors $300 for articles accepted after peer review.

In theory, that would work just fine for everyone. Libraries would still subscribe to most everything, and individual researchers would subscribe to the few first-class journals necessary for staying on top of their fields. Publishers and authors would both be pleased that more people, the world over, will have access to their work. The public would be happy because the more information in play, the better. Authors presumably might chafe at paying to publish their work, but that system already exists today in certain instances. In practice, authors would simply shift the cost of publishing into their government grant applications, and, in any event, everyone would be spending less money on subscriptions, freeing up pocket change. According to Roberts, since the government also often funds the libraries that pay for and support scientific journals, the public already "pays for the research to be carried out, pays for it to be published, and ultimately pays to read it ... There is no new money needed, only a redistribution."

And, in fact, a number of journals have decided to transition to this model, and have had some success. Elizabeth Marincola is executive editor of Molecular Biology of the Cell, a magazine available through PMC six months after it mails to subscribers. According to her: "The big fear was that going to PubMed Central would undermine subscriptions. In our case, though, access and interest peak very quickly after the issue is released. And our subscriptions increased last year."

Roberts is editor of Nucleic Acids Research, another of the handful of publications currently available on PMC. But his journal didn't exactly leap into the new world. When asked why his journal decided to play, Roberts deadpans: "When I offered to resign, the publisher decided to do it."

Pub Crawl

Fully implemented, PMC's agenda would surely force some consolidation in scientific publishing. It would also take a bite out of some of the larger publishers that have done quite well as the price of scientific journals has increased, on average, by 170 percent over the past decade and a half.

But opponents of PMC also voice serious practical concerns, and there are dangers in consolidating such an essential resource. An angry and irrational Congress could zero out funding for PMC one year. The technology used at NIH could fail or be surpassed. And how much control should the federal government have over science in any event? One editor of a scientific journal emailed that the government "already controls the funding of most biomedical research. They regulate most aspects of it. They manage the large central databases like [the human genome database]; and now they'd like to manage the literature. How much of this is Science for the People and how much is funktionlust."

Time will tell whether some cranky Congress axes funding, but it's not something to bet on. PubMed Central currently costs about $3 million a year to maintain, about one 60th of one percent of NIH's annual budget, and it would only cost about $100 million to post extensive archives going back 30 years. In addition, NIH may well ultimately stick the World Health Organization (WHO) or another international body with the bill. That may not be a bad idea, given that Swedes and Zimbabweans will have the same access to PMC as Americans and that PMC should actually have a proportionally greater impact on Third World medicine than First World. The Soros Foundation has already begun work with six major publications and the WHO to make up to 1,000 journals freely available online across the developing world.

Opponents also fear that PMC could dilute quality control. By its nature, PubMed Central would be an equal-opportunity database. A study on immune systems subjected to rigorous peer review and reexamination by the prestigious Journal of Immunology would not receive priority over a similar article by a slipshod competitor, or even a PubMed-compliant journal published by a company standing to make millions of dollars from the subject of the article. Right now, slipshod journals have to raise the money to send out issues and convince libraries to subscribe. With PMC, they'd only have to send their data across the Net. PMC only accepts peer-reviewed work, but it also currently only requires that a journal have two federally funded researchers on its board--a very low bar for an industry that has repeatedly found ingenuous ways to co-opt science and scientists. According to Dr. Sidney Wolfe of the activist group Public Citizen: "The whole thing begs for controls as strong as possible. Otherwise you'll get thinly disguised advertisements for drugs."

The immediate counter to that concern is that the horse has already bolted out of that barn. Corporations and snake-oil salesmen have flooded the Internet with their news. PubMed Central may be imperfect, but the information it puts out will almost certainly exceed the quality found online today, and may not be that much different from the quality of scientific publishing out there. If a 70-year-old man decides to devote himself to researching his health online, maybe he'll then come to his doctors with questions about the latest research in the New England Journal of Medicine, not with a jar of lamb placentas or other potentially dangerous nutritional supplement so easily found online. Nor will hucksters be able to fool the most serious scientists using PMC. PubMed Central's director, David Lipman, says, "We are going to be improving the proportions of good to bad information, and canceling out some of the charlatans selling useless and dangerous stuff."

Indeed, scientists in other fields have used the Internet to create new forms of peer review, some just as effective as the old methods. With Paul Ginsparg's physics site developed at Los Alamos Laboratories, arXiv, for example, no one waits six months from publishing date, or even six months from the time he completes his work. Researchers post their data online almost as soon as they have it. Other physicists examine the data, comment, post their own comments, reexamine the data sets, and the article and data evolve over time. MIT Press has set up a virtual community called CogNet for cognitive and brain scientists. The scientists subscribe to the service, discuss the hottest current research in the field, post their comments, and watch as the discussions gradually evolve. Quality control is essentially self-imposed and not unlike the system used on several commercial sites, such as eBay or Slashdot, where users are graded by their peers or site editors and lose points if they prove untrustworthy.

Free Bird

PMC represents one more example of a growing battle over information freedom. Our technological boom has increased both the value and availability of information, creating a tension between those who want to hold on to it, and those who want to give everything away free. Computer programmers have clashed over supporting Linux--the operating system created by a loosely organized band of unpaid programmers that competes with Microsoft Windows and is available for free over the Internet (See "Reboot," March 2000). The medical community has fought this battle over GenBank--the government's free database of gene sequences created out of the human genome project--and is still fighting over patents. Many activists want the patent system opened up so that small drug companies can develop generic lifesaving drugs. Drug companies counter that weakening patent protections would take away their financial ability to find and create new drugs.

But of all these battles, the one over PMC seems to be the one where freedom really ought most to win out. Unlike with Linux and Microsoft, the government can change the system because the government already runs the system. Moreover, opening up information doesn't pose the debilitating threat to innovation that it might with patents. Some of the big publishing houses could still survive with PMC, and in any event, Elsevier provides much less value to scientific publishing than Eli Lilly, for example, does to the creation of pharmaceuticals. Donald Kennedy, editor in chief of Science magazine and by no means a radical on this issue, compares Elsevier to a maquiladora: "It's a well-capitalized European publisher, taking United States copy, adding little value, and then selling it back here at huge profits."

We should hope that the boycott succeeds in making PMC universal, and that peer review and quality-control mechanisms evolve, whether through the same old journals and a greater online presence, or maybe through models similar to what CogNet and arXiv use today. New ways of opening up information will also follow. Even the discussion of PubMed Central has spawned the creation of alternatives that the publishing industry considers more palatable, for example CrossRef, a pay-as-you-go archive of 1,200 journals.

There's a lot at stake: whether one is an aspiring Watson or Crick at a university without a first-rate library, a doctor in a meningitis laboratory in Zimbabwe, or even just a participant in a Johns Hopkins study on hexamethonium. With so little to lose, the more information the better.

Research Assistance by Alex Walker

NICHOLAS THOMPSON is an editor of The Washington Monthly.