In his home between two Indian pueblos, Dr. Paul Ginsparg, 42, built an online archive system that is changing the face of scientific discourse.Every day, tens of thousands of physicists in more than 100 countries point their Web browsers to a site maintained by the Los Alamos National Laboratory in New Mexico. There, they peruse research papers that have been posted since their last visit, trolling for new work in, say, quantum cosmology or statistical mechanics, and send E-mail to one another based on the day's postings. Many of them no longer read print journals. They don't need to. The Los Alamos archive -- a vast repository of physics preprints, or E-prints -- is the primary means for today's physicists to exchange information.
The archive is the product of nearly seven years of effort on the part of Paul Ginsparg, a 42-year-old physicist who believes that the Internet is ushering in the end of paper-based communication among research scientists. In particular, Dr. Ginsparg's system is a direct challenge to the hallowed 200-year-old practice of peer review, the process by which a scientific paper is evaluated by colleagues before it is published.
"Traditional peer review has always been an awkward compromise imposed by the inadequacies of the paper medium," he said. The Los Alamos archive's following has fueled a debate among scientists from a cross-section of disciplines. Some view Dr. Ginsparg's achievement as intellectual history in the making -- a harbinger of what lies in store for all scientific discourse. Others eye it warily, calling it an anomaly in science, well suited to physics perhaps, but not easily adapted to other, larger disciplines like medicine.
Peer review upholds a central tenet of scientific publishing: only after careful scrutiny by experts who vouch that a fellow scientist's work is worthy -- that it contains no errors and repeats no earlier work -- should a paper be published.
While Dr. Ginsparg's archive has no traditional peer review per se, scientists who post their papers to the Los Alamos site typically revise their submissions in response to direct feedback from others. "Subsequent revisions frequently benefit as much or more from this feedback as from any conventional referee process," said Dr. Ginsparg in an interview that was, at his insistence, conducted entirely by E-mail.
The site (http://xxx.lanl.gov) also employs what Dr. Ginsparg calls version control, meaning that any earlier version in a series of revisions can be viewed.
Dr. Ginsparg, highly regarded as a high-energy particle theorist, did not set out to build a digital Alexandria. He started it in 1991 to speed the distribution of drafts to other theoretical physicists.
A new game with no referees
He wrote the software for the original system in his spare time, building much of it via a connection between the Los Alamos Lab and his home, 20 miles away "in the middle of nowhere between two Indian pueblos." For years the entire system resided in a computer under a table in Dr. Ginsparg's office. When noise from the fan and disk drive grew too annoying, he moved it down the hall.
Since then, Dr. Ginsparg has added 30 subdisciplines to the site and separate sections for mathematics and so-called nonlinear sciences like evolution. Computer scientists will soon join, too. More than 35,000 people visit the site each day.
A colorful presence in science, Dr. Ginsparg is known for thinking against the grain. He argues that traditional peer review -- controlled by scientific journals -- will evolve into a system in which refereeing is separated from publishing. He does not disguise his disdain for scientific journals, many of which he believes are unnecessary intermediaries, slow-moving and run by profiteers.
Some scientists believe that electronic systems like Dr. Ginsparg's are actually changing the scientific process itself, especially considering the months that usually pass between the time a paper is submitted to a print journal and its publication.
Moreover, say some, the worldwide readership of the archive quickly discovers errors and helps eliminate research duplication.
Skeptics of Internet publishing say that its long-term archival ability has not been established. They also worry about having a system dependent on one person or machine.
Dr. Ginsparg, who receives funding from the National Science Foundation and the Department of Energy for the archive, has taken steps to allay such fears. The information on the archive is duplicated, on more than a dozen Web sites, and the American Physical Society is planning to incorporate the archive into its publishing operation by way of crosslinks and references.
Other scientists have voiced concern about the quality of work published electronically. But the Los Alamos system stands in contrast to freewheeling Internet news groups where anyone can muse on quantum mechanics without so much as freshman calculus. "We try to maintain the ivory tower isolation to protect us from random passers-by interrupting our seminars," Dr. Ginsparg said. In practical terms, that means that submissions from, say, aol.com or hotmail.com get bounced.
As a result, the quality of submissions to the archive remains consistently high. "There seems to have been no serious problem of people submitting inappropriate matter, or papers so badly written as to chiefly constitute a waste of the readers' time," said Charles Bennett, a physicist at the International Business Machines Corporation. "I can't say that of the mail I get in my mailbox or my E-mail inbox."
Another point of contention the archive has brought into sharp focus is prepublication. The journals' practice of restricting dissemination of findings prior to publication can run counter to the authors' interests, said Dr. Ginsparg.
Now even the most stalwart journals are making concessions. While editors at the prestigious Nature do not encourage posting a paper to a site like Dr. Ginsparg's, they do not forbid it. "E-print servers to my mind are an extension of an idea," said Philip Campbell, Nature's editor in chief. "It's just a form of intra-scientific communication."
But while Dr. Ginsparg's system might work well for physics, its suitability for other fields is open to question. "Physics is a small community of people, and what they put out on their Web site is not going to affect the health of thousands or millions of people," said Jerome Kassirer, editor of the New England Journal of Medicine, which publishes 10 percent of the papers it receives.
As for peer review, Dr. Kassirer said, "like democracy it's the only thing we have that really does work."
Not surprisingly, Dr. Ginsparg disagreed. "The question is whether we can do without peer review as organized by multinational commercial publishers," he said. "If the answer is yes, the likely benefit is a far more efficient and far more functional system, better adapted to the needs of researchers."
Alan Guth, a well-known cosmologist at Massachusetts Institute of Technology and a regular reader of Dr. Ginsparg's archive, said, "The journals are now playing the role of being historical arbiters of truth rather than real disseminators on a day-to-day basis."
If Dr. Ginsparg is right, it could spell the demise of the journals themselves, relegating them "decorative paperweights."
William Loughlin, physical sciences bibliographer at the University of Georgia, said of Dr. Ginsparg, "He makes the publishers very, very nervous."
Other experiments with scientific communication and peer review are beginning to emerge. The Medical Journal of Australia is trying Web-based peer review. Biomedical researchers at the University of Pittsburgh are conducting the first all-electronic peer review under a National Institutes of Health grant. And a group of university administrators has proposed a decoupling system for peer review that would have an independent board of experts certify a paper, free of ties to any journal.
Still, some of Dr. Ginsparg's admirers are disappointed that, after seven years, his system remains the exception. Ever concerned about promotion, tenure and their general reputation, even most of the scientists who now read the archive continue to submit their work to journals after posting it to Los Alamos.
"Sooner or later someone is going to be shrewd and prophetic enough to realize that Paul has quietly done something absolutely monumental," said Stevan Harnad, director of the Cognitive Sciences Centre at Southampton University in England, who has built an archive modeled after the Los Alamos system. "When the historians write the history of it all, Paul Ginsparg will certainly get the full credit for having shown the way, not just by pointing, but by actually constructing the ultimate solution."
Copyright 1998 The New York Times Company