You have to wonder at some people. I have been wondering at Jon Sudbo, a Norwegian scientist who published a paper in The Lancet in 2005 showing that a certain class of painkillers cut the risk of oral cancer. Sudbo, it turned out, made the whole lot up. And he was astoundingly dim in the way he went about inventing his 908 patients: he gave 250 the same date of birth.

As I learnt at a terrific conference in London last week, hosted by the charity Fraud Advisory Panel, there are many more Sudbos out there but scant means of spotting them. The handful who are found must be a tiny minority, said Philip Campbell, editor-in-chief of Nature. And so, he says, we need to consider “going the extra mile” to find them. He is considering whether some studies, especially ones that make headlines, should be replicated before going to press.

Science operates on an assumption of honesty – raw data are rarely scrutinised by either institutions or journals, and academics are encouraged to work independently. Rogue researchers feed off this culture of trust: busy superiors and colleagues often sign off research papers and grant applications without reading them. Fame ensues and grants and citations roll in.

And so it becomes hard to “out” a suspect. Do you snitch to your head of department, for example? To your vice-chancellor? Might he or she wish to conceal an issue that could make the institution look culpable? If the person moves and you divulge your suspicions to his new employer, can you be sued?

One solution is to make whistleblowing easier. On Friday the Research Integrity Office, a panel set up last year to promote good practice in biomedical research, launched a confidential hotline for the reporting of misconduct in universities, industry and the NHS (0844 7700644). About 1 per cent of clinical trials are thought to be suspect. This can distort the literature and put patients at risk.

It is a useful step but a modest one: it does not deal with bad behaviour in the physical sciences. And the onus is still on the host institution to investigate and punish. As Dr Campbell told me, some institutions take this responsibility more seriously than others. Woo Suk Hwang, the South Korean biologist who falsely claimed to have cloned a human embryo and extracted embryonic stem cells from it, was brought down chiefly by his own university. Others close ranks.

The conference brought a provocative contribution from Nicholas Steneck, a scientific fraudbuster from the University of Michigan, who pointed out that while plagiarism is undesirable, it may do less harm than the commoner practice of altering data analysis methods to achieve a desired result.

Professor Steneck asked: “What does plagiarism do to the literature? Not very much – as long as the plagiariser is accurate.” And provided, of course, that the person whose work you’re copying has higher standards of integrity than you.

Times Recommends

• A dose of defeat is just the medicine
• Protecting ourselves from Eurocreep
• After the summer of LUV, it'll still hurt