FOR the first time in a long while I have plenty of empty shelves in my office. I have just thrown out all the scientific journals that I鈥檝e collected over the past 30 years. A little rash of me? Not at all. They had long passed their usefulness.
For a start, back numbers are available on the Web. More importantly, I don鈥檛 read new journals as much as I used to. This is not because I am less eager to learn their contents. Rather, most of what鈥檚 in them has already been published electronically, in particular on the famous physics preprint server () started some years ago by Paul Ginsparg, now at Cornell University in New York State. On a typical day I scan about 40 abstracts on this website.
The Web offers many advantages for disseminating new science. It is more democratic and more efficient. Electronic preprints can reach almost everyone at the same time, and more people read them. They have a (perhaps spurious) aura of immediacy. An exciting idea from a student in, say, India can trigger a flood of emails within 24 hours from scientists around the world. With printed journals such a response could take a year.
Advertisement
Electronic publishing may render the traditional printed journal redundant. But what will it mean for the way papers are refereed? Most scientists accept peer review as an effective way to filter out shoddy work and unmerited claims. Indeed, most of us believe it is crucial that 鈥減roper鈥� peer-reviewed journals should continue, even if only in electronic form. But even then, they will play a diminishing role in the way scientists publicise their work and learn about that of others. Commercial pressures or intense academic rivalry are already leading to many discoveries being widely publicised and dissected before they have been formally refereed.
This need not be at the expense of excellence. Refereeing by learned journals is not the only way to ensure quality control in science. Indeed, this time-honoured process has suffered some much-publicised failings recently: both Nature and Science published several papers from researchers at Lucent Technologies鈥� Bell Labs that have since been judged worthless. With electronic publication, peer review is supplemented by more open and effective criticism. Preprints get far wider scrutiny than journal papers. I tend to receive more feedback from colleagues who read my preprints than from official referees.
What about a system that allowed people to comment on papers online, rather like the informal book reviews on Amazon.com? Authors could compete for 鈥渇ive star鈥� endorsements from leading scientists in their field. That would be a far more satisfying accolade than getting a paper accepted in a high-prestige journal.
Some people argue that this would lead to an uncontrolled flood of bad papers. But the effect could be the reverse. At the moment there is an incentive to publish lots of papers just above the threshold of acceptability. If there were no merit in mere publication, people might write less.
But the most important change would be to the way science is debated. Scientific claims must be challenged. Traditionally this takes place in private, between the authors, the journal editors and the referees. The growth of online publication should make it possible to open up these debates.
One of the most publicised departures from scientific norms happened in 1989 when Martin Fleischmann and Stanley Pons claimed to have achieved 鈥渃old fusion鈥�. This would have been one of the greatest advances since the discovery of fire. But doubts soon set in. Many tried to reproduce their results, but all failed. Within a year the general consensus was that cold fusion was a fantasy.
The episode carries an important lesson. The fiasco did no great harm in the long run, except to the personal reputations of those involved. But suppose a claim as extraordinary as theirs were made by scientists in a military or commercial laboratory who managed to convince their bosses of its unprecedented economic and strategic importance. A massive secret research programme would soon get under way, consuming huge resources and shielded from open scrutiny.
Something very like this happened in the 1980s, when the US Department of Energy鈥檚 Lawrence Livermore National Laboratory launched a secret programme to produce X-ray lasers 鈥� intense 鈥渞ay guns鈥� triggered by a nuclear explosion that could destroy incoming enemy missiles. Independent experts were almost uniformly scathing about this scheme, but it was the brainchild of Edward Teller and his prot茅g茅s. Working in a closed environment, they could commit vast Pentagon resources to it. If today similarly well-connected researchers were to devise a new source of energy, one can well imagine demands for a crash programme to bring it to fruition.
Without formal peer review, access becomes ever more important. Many of us won鈥檛 go to the wall to defend paper journals. But we shouldn鈥檛 compromise on the need for openness.