Your Opinion pieces propose that research misconduct could be prevented either by financial incentives for teaching research integrity or by informal intervention (Nature 466, 436–437 2010 and Nature 466, 438–440; 2010). Weak regulations and admonitions are unlikely to deter a prospective fraudster, as they are easily dodged.

The complex, idiosyncratic and ephemeral nature of much research encourages misconduct. Most instances of cooked data go undetected. Formal accusations to institutional officials are undermined by risk-management policies that try to minimize evidence of misconduct. And reasoned suspicion of incredible published findings is countered by the tendency of journals and leaders in the field to promote spectacular results.

Spotting fraud in a publication depends on a chance finding of identical data in different experiments. Even blatant and extensive incidents — as in the case of plastic transistors (see E. S. Reich Plastic Fantastic Palgrave Macmillan; 2009) and of reactome chemistry ( see J. Travis Science 327, 22–23; 2010) went undetected by colleagues, reviewers and journal editors.

To deal with research misconduct, we need to uncover the extent of the problem and the factors encouraging it. A common attitude is that misconduct is rare and has little impact because science is self-correcting. Such complacency produces an environment in which fraudsters can flourish.