Most researchers agree that peer review is the least imperfect way of upholding the quality of scientific publications. But those who administer it also have to cope with, and attempt to solve, the problems that peer review gives rise to.

One such problem is misconduct. The peer-review process depends on trust, and the great majority of reviewers are trustworthy, but there are occasions when that trust has been abused by a referee. One of the worst that Nature has experienced, some years ago, was when a referee obstructed a paper, and used its information indirectly to obtain materials from the author to complete competing work of his own, which he then promptly published, thereby scooping the original. (The referee has not been used again.)

Like most journals, Nature requests that referees disqualify themselves from refereeing a paper that is directly competitive with their own work. We also request that referees either keep the paper strictly to themselves or consult, also in confidence, a close colleague whom they identify to us. We use two and often three referees, and do what we can (which we believe is quite a lot) to spot phoney arguments against a manuscript and to sidestep undue delay.

But there is only so much that Nature or any journal can do. Take confidentiality: as can be seen in a survey of the problems (see page 102), scientists vary in their attitudes to the principle of strict confidentiality. Some freely consult colleagues to an extent that others (including some authors) would find horrifying. And it is not unknown for a Nature editor to visit a lab and see a manuscript sent in confidence left casually on display on the group coffee-table.

Even more concerning is a discernible growth in cases where referees sit on manuscripts, or are subsequently shown (or strongly suspected) to have plagiarized them, or used them as a stimulus to rush their own, sometimes less-developed versions, into print in a competing journal. On the rare occasions where the offence is undeniable, Nature can take sanctions, for example by contacting the offender's employer.

Misconduct in peer review is very infrequent. But in the most competitive areas of biology — molecular biology, in particular — it is, in Nature's experience, no longer unexpected. In seeking to address this regrettable situation, it is important to examine underlying pressures.

The particular competitiveness in molecular biology appears to stem from a number of factors, not the least of which are the rampant egotism in the upper echelons of the field and the urgent need to publish if postdocs and contract researchers are to obtain grants. Furthermore, the route from basic research to commercial exploitation is a particularly short and direct one. Duplication of effort is yet another factor. The risk that years of painstaking work will be reduced to insignificance through being scooped by competitors seems significantly greater in molecular biology than in other disciplines. This is partly because of the inexorable pressure on researchers to attack problems guaranteed to yield quick and safe scientific returns.

With the best of intentions, science publishers are now inadvertently exacerbating the situation. With every encouragement from researchers, more and more journals (including Nature and its related titles) are moving in the direction of electronic publication ahead of print. This can be good for science, and especially so where the research has immediate positive implications for human health, for example. But everyone suffers if the practice is allowed to encourage corner-cutting in research and to further stimulate bad behaviour.

Perhaps it is time to review the concept of the scientific scoop. Many journals, including this one, are tough on authors of papers under revision whose results are scooped by others. But papers reporting similar results published weeks apart will all be found by an electronic search of the literature, and the later publications may be of higher quality by virtue of the additional time taken. Maybe there is a better balance to be struck between the recognition of quality and of priority than is currently the norm.