Begin forwarded message:
Date: October 19, 2012 4:40:35 AM EDT To: Randy Read <[email protected]<mailto:[email protected]>> Subject: Re: [ccp4bb] PNAS on fraud This thread has been quite interesting to me. I've had a long interest in scientific fraud, which I've generally held to be victimless. While that view is unsupportable in a fundamental sense, I feel strongly that we should understand that error correction costs exponentially more, the smaller the tolerance for errors. In protein synthesis, evolution has settled on error rates of ~1 in 4000-10000. Ensuring those rates is already costly in terms of NTPs hydrolyzed. NASA peer review provided me another shock: budgets for microgravity experiments were an order of magnitude higher than those for ground-based experiments, and most of the increase came via NASA's insistence on higher quality control. Informally, I've concluded that the rate of scientific fraud in all journals is probably less than the 1 in 10,000 that (mother) nature settled on. I concur with Randy. Charlie On Oct 18, 2012, at 2:43 PM, Randy Read wrote: In support of Bayesian reasoning, it's good to see that the data could over-rule our prior belief that Nature/Science/Cell structures would be worse!
