How to be better at fraud

Sat 06 Feb 2010 11:06 AM

We often assess claims based on plausibility of style and content. In writing about Wikipedia, I argue that these assessments can be frustrated by community editing. The implausible details can be taken out of false accounts, making the falsity harder to detect. Some people respond to my argument by denying that this happens.

Reading Eugenie Samuel Reich's Plastic Fantastic, I bumped into a similar phenomenon. Reich is a science journalist, and the book is about fraudulent science. Her claim is that peer review does not do an especially good job of catching deliberate fabrication. Moreover, scientists who perpetrate fraud often exploit reviewers' comments and questions in order to make their fabrications more plausible. Reich writes:

Not only in there no guarantee that a thorough review process will detect a false claim, but even more disturbingly, a thorough review may do little more than reveal to authors what changes they need to make in order to turn a false claim into a more plausible scam.[p. 122]
The parallel with Wikipedia is not precise, but in both cases conscientious but imperfect editorial oversight results in public versions which are more plausible false accounts than the original submissions.

Scammer scientists exploiting this can publish more plausible scam papers than they could have otherwise. Yet one might hope that, although this helps fraudulent papers on the timescale of months, fraudulent research programs will still be uncovered in the course of just a couple of years. The parallel hope for the Wikipedia is that false claims will be corrected eventually.

So the hope is that fraud burns brighter by exploiting peer review but will still burn out in relatively short order. Consistent with this is the fact that none of the cases of fraud which Reich describes have gone undetected for more than a few years, and each was discovered in time to ruin the scientists responsible. Yet that may just be because she can only report scientific fraud which was ultimately detected.

[html]<div align="center"><iframe src="" style="width:120px;height:240px;" scrolling="no" marginwidth="0" marginheight="0" frameborder="0"></iframe></div>[/html]